• Microsoft sees enterprise AI moving from adoption to transformation. 

• Microsoft 365 Copilot has surpassed 30 million paid seats, showing that enterprise AI is entering a broader scale-up phase. 

• AI agents are starting to participate in actual work execution, not just provide suggestions. 

• The future success metric will shift from “how many people use AI” to “how businesses redesign workflows around AI.” 

• Hong Kong organizations should focus on governance, workflow design, and organizational readiness, not only productivity gains. 

Over the past two years, generative AI has entered the workplace faster than many businesses expected. From content creation and data analysis to meeting summaries and knowledge search, more organizations are now embedding AI into daily operations. 

But as AI adoption matures, an important question is emerging: should businesses still measure AI success mainly by license counts, usage rates, or hours saved? 

Microsoft’s latest perspective suggests that the next measure of AI momentum is not simply adoption, but whether work itself is being redesigned and transformed. In other words, the next phase of enterprise AI should focus on how work changes, not only how many people use AI. 

Video Source: Microsoft 365 Copilot: Work, transformed — Microsoft 

According to the Microsoft 365 Blog, Microsoft 365 Copilot has surpassed 30 million paid seats, reflecting how enterprise AI is moving beyond early experimentation and into broader business adoption. 

What is Microsoft 365 Copilot?
Microsoft 365 Copilot is an enterprise-grade AI assistant integrated into the Microsoft 365 platform. It connects with Microsoft Graph and organizational data to help employees search knowledge, analyze information, and complete work tasks. 

Microsoft emphasizes that AI is evolving from a passive tool into an active participant in work. Employees are no longer only asking AI questions; they are increasingly directing AI agents to handle tasks, projects, and processes. 

At the same time, more organizations are redesigning workflows around AI rather than simply adding AI to existing processes. This is why Microsoft frames the next milestone as transformation, not just adoption. 

Image Souce: The next measure of AI momentum is work transformed 

Many organizations begin their AI journey by focusing on adoption metrics: how many users have access to AI tools, how often they use them, and whether they can save time on common tasks. These indicators are useful, but they do not fully capture business impact. 

AI transformation goes one step further. It asks whether business processes, decision-making models, and employee workflows are being redesigned to take advantage of AI. For example, a sales team may use Copilot not only to summarize meetings, but also to prepare account insights, draft follow-up emails, identify next-best actions, and support pipeline reviews. 

For Hong Kong businesses, this distinction matters. Many SMEs and enterprises are already facing pressure to improve productivity, reduce manual work, and make better use of internal knowledge. AI can support these goals, but only when it is connected to clear business use cases and supported by proper governance. 

Image Souce: The next measure of AI momentum is work transformed

Microsoft highlights several major shifts in the way organizations are using AI at work. First, AI is becoming more deeply embedded in daily productivity applications, including email, meetings, documents, spreadsheets, and collaboration tools. This allows employees to work with AI in the flow of work rather than switching between separate systems. 

Second, AI agents are beginning to handle more task-oriented work. Instead of simply answering questions, agents can help coordinate processes, retrieve information, generate drafts, and support multi-step workflows. This opens the door to more scalable automation across departments such as sales, marketing, finance, HR, IT, and customer service. 

Third, organizations are starting to rethink roles and responsibilities. As AI takes on repetitive or information-heavy tasks, employees can spend more time on judgment, creativity, customer engagement, and strategic work. This is where AI moves from being a productivity tool to becoming part of the operating model.

For Hong Kong organizations, AI transformation should not be treated as a purely technical project. It is a business change initiative that requires alignment across leadership, IT, operations, and end users. 

Organizations that want to move beyond basic AI adoption should start by assessing their Microsoft 365 readiness, identifying high-impact Copilot use cases, and building an adoption roadmap that balances productivity, security, and governance. 

1. Start with high-value business scenarios.
   Organizations should identify use cases where AI can create measurable business value, such as reducing time spent on reporting, improving sales follow-up, accelerating proposal creation, enhancing customer service, or supporting internal knowledge search.
2. Strengthen data readiness and governance.
   AI outcomes depend heavily on the quality, accessibility, and security of business data. Before scaling Copilot or AI agents, companies should review permissions, information architecture, data protection policies, and compliance requirements.
3. Support employee adoption and change management.
   AI transformation requires people to change how they work. Training, practical use-case sharing, prompt guidance, internal champions, and ongoing support are essential to help employees build confidence and develop new habits.
4. Move from individual productivity to process transformation.
   The biggest return on AI will come when organizations use it to redesign workflows across teams. This may include automating routine handoffs, shortening approval cycles, improving response time, or connecting data across departments.

• AI value should be measured by business outcomes, not only adoption statistics. 

• Copilot and AI agents can support both individual productivity and cross-team workflow transformation. 

• Data governance, security, and permissions must be addressed before large-scale deployment. 

• Successful AI adoption requires training, change management, and leadership alignment. 

• Organizations that redesign work around AI will be better positioned to improve efficiency, responsiveness, and competitiveness. 

Microsoft’s latest message is clear: the next chapter of AI is not just about putting AI tools into employees’ hands. It is about transforming how work is structured, executed, and measured. 

For Hong Kong businesses, Microsoft 365 Copilot and AI agents offer a practical path to improve productivity, unlock organizational knowledge, and redesign workflows. However, the most successful organizations will be those that combine technology deployment with governance, adoption planning, and business process change. 

1. Is Microsoft 365 Copilot only for large enterprises?
No. While large enterprises may have more complex deployment needs, SMEs can also benefit from Copilot when they start with clear use cases such as email drafting, meeting summaries, document creation, reporting, and knowledge search. 

2. What should businesses prepare before deploying Copilot?
Businesses should review data access permissions, security policies, information management practices, user readiness, and priority business scenarios before broad deployment. 

3. How can companies measure AI success?
Beyond usage and time savings, companies should measure business outcomes such as faster response times, improved work quality, reduced manual effort, shorter process cycles, and better employee experience. 

4. Why is governance important for AI transformation?
AI tools rely on business data. Without proper governance, organizations may expose sensitive information, duplicate inaccurate content, or generate inconsistent outputs. Good governance helps ensure AI is used securely, responsibly, and effectively. 

Source: The next measure of AI momentum is work transformed | Microsoft 365 Blog

• Microsoft Sales Agent is an AI agent within the Microsoft 365 Copilot ecosystem, designed to support sales teams in customer preparation, opportunity follow-up, CRM updates and post-meeting actions. 

• It connects CRM data, customer interactions, Outlook, Teams and Microsoft 365 work context to help sellers understand customers faster. 

• Its main business value is reducing administrative work, improving sales follow-up and helping teams spend more time on high-value customer engagement. 

• For Hong Kong businesses, it is especially relevant for B2B sales teams with long sales cycles, complex accounts and frequent customer follow-ups. 

Microsoft Sales Agent shows how AI is moving from general productivity support into role-specific business workflows. For sales teams, its value is not to replace CRM, but to connect CRM, Outlook, Teams, meeting notes and customer interactions so sellers can understand customers, move opportunities forward and complete follow-ups faster. 

Many sales teams already use CRM, but customer information is often scattered across systems. This creates extra time spent searching for records, updating notes and preparing follow-ups instead of focusing on customer conversations and deal progress.

Image Source: Sales Agent is now generally available: Bringing customer and deal intelligence into the flow of work 

Microsoft Sales Agent is a role-specific AI agent for sales teams. It helps sellers bring together CRM records, customer interaction history and Microsoft 365 work data to prepare for meetings, manage opportunities and complete follow-up tasks more efficiently. 

In simple terms, CRM stores sales information. Sales Agent helps turn that information into usable customer context, suggested actions and workflow support. 

1. Customer & Account Intelligence
   Sales Agent can summarize account background, recent interactions, key stakeholders and relationship history. This helps sellers prepare faster before customer meetings without manually checking CRM, emails, Teams messages and meeting notes one by one. 
2. Opportunity & Deal Insights 
   For pipeline management, Sales Agent can help summarize opportunities, highlight stalled deals and identify accounts that may require follow-up. This is useful for B2B teams managing multiple customers and longer sales cycles. 
3. Meeting & Conversation Intelligence 
   After meetings, Sales Agent can help generate summaries, capture decisions, suggest follow-up actions and support CRM updates. This reduces manual admin work and lowers the risk of missing important customer information. 
4. CRM Update & Administrative Work Reduction 
   Sales Agent can reduce CRM-related admin by helping sellers update records, organize notes and manage follow-ups through natural language. This makes CRM easier to maintain and more closely aligned with daily sales activity. 
5. Sales Productivity & ROI 
   From an ROI perspective, the value is not just saving a few clicks. Sales Agent helps sellers spend less time on meeting preparation, CRM updates and follow-up drafting, and more time on customer engagement and deal progression. 

Primary Use 

Microsoft Sales Agent is designed for sales workflows and opportunity management. It helps sales teams manage customer context, sales activities, CRM updates and follow-up actions more efficiently. Microsoft 365 Copilot, on the other hand, is broader and supports general workplace productivity across emails, documents, meetings and collaboration. 

Target Users 

Sales Agent is mainly built for sales teams, including sellers, account managers and sales managers who need to manage customer relationships and pipeline activity. Microsoft 365 Copilot is designed for all employees who want to improve everyday productivity across Microsoft 365 apps. 

Core Value 

The core value of Sales Agent is to improve deal follow-up, sales execution and customer engagement. Microsoft 365 Copilot focuses on improving individual and team productivity by helping users work faster with information, content and collaboration tasks. 

Relationship between Sales Agent and Microsoft 365 Copilot 

The two tools are complementary. Microsoft 365 Copilot helps employees work more efficiently across documents, emails, meetings and collaboration. Sales Agent applies AI more specifically to sales workflows, customer context, opportunities, CRM updates and follow-ups.

Scenario 1: Meeting Preparation 

Image Source: Moving sales and service organizations forward with agentic CX and Microsoft 365 Copilot 

Before a customer meeting, Sales Agent can summarize previous meetings, recent emails, open opportunities and key stakeholders. Sellers can prepare in minutes instead of manually collecting information from different systems.

Scenario 2: Opportunity Prioritization 

Image Source: Moving sales and service organizations forward with agentic CX and Microsoft 365 Copilot 

Sales managers can use Sales Agent to identify stalled deals, review customer engagement and prioritize accounts that need attention. This helps teams focus on opportunities with higher impact or higher risk. 

Scenario 3: Post-Meeting Follow-Up 

After a meeting, Sales Agent can help draft a summary, recommend next steps, update CRM and prepare a follow-up email. This improves consistency and reduces the chance of missing important actions. 

u003cspan data-contrast=u0022noneu0022u003eDocument understanding is one layer of AI readiness. Before enterprises deploy AI at scale, they also need to understand whether their cloud environment, application landscape, data sources and operational processes can support the next step.u003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003ernrnu003cspan data-contrast=u0022noneu0022u003eThrough recent Azure POV and cloud migration assessment work, Superhub has observed that many Hong Kong enterprises want to explore AI, Copilot or automation, but are still unclear about their current IT readiness, migration priorities and possible deployment roadmap.u003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003ernrnu003cspan data-contrast=u0022noneu0022u003eAzure POV provides a practical starting point by helping organizations assess existing infrastructure, applications, and operational processes, then identify whether they should begin with cloud migration, modernization, document understanding, RAG, u003c/spanu003eu003ca href=u0022https://www.superhub.com.hk/blog/microsoft-365-copilot-adoption-services-business-value/u0022u003eu003cspan data-contrast=u0022noneu0022u003eCopilotu003c/spanu003eu003c/au003eu003cspan data-contrast=u0022noneu0022u003e Extension, or AI Agent scenarios.u003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003e

Agentic sales is about moving AI from answering questions to helping complete workflows. Sales Agent reflects Microsoft’s direction towards AI tools that can support real business processes, not just generate content. 

Expert Perspective: The Biggest Challenge Is No Longer AI 

SUPERHUB believes the biggest challenge is not the AI technology itself, but data quality and sales process maturity. If CRM records are incomplete or sales stages are unclear, Sales Agent may still produce summaries, but its recommendations and follow-up actions will be less reliable. 

Professional Services 

Professional services firms often manage complex client relationships and require significant preparation time before meetings. Sales Agent can support account summaries and follow-up tracking, helping teams prepare faster and respond to clients more efficiently. 

Financial Services 

Financial services organizations usually handle large volumes of customer data and relationship history. Sales Agent can help summarize customer intelligence and relationship context, enabling better client engagement and more informed conversations. 

Trading & Distribution 

Trading and distribution companies often manage many customers and scattered opportunities at the same time. Sales Agent can help with pipeline follow-up and prioritization, improving follow-up efficiency across a large customer base. 

Technology & IT Services 

Technology and IT services companies often deal with fast-changing customer needs and upsell opportunities. Sales Agent can provide account intelligence and next-step suggestions, helping teams strengthen account development and identify new opportunities more effectively. 

Before adopting Sales Agent, organizations should review whether their CRM data, sales stages, customer ownership rules and Microsoft 365 usage are ready for AI-assisted selling. 

• Clean and complete CRM records 

• Clear sales stages and follow-up rules 

• Consistent use of Outlook, Teams and meeting notes 

• A pilot group to validate workflows before wider rollout 

1. What is Microsoft Sales Agent? 
   Microsoft Sales Agent is an AI agent for sales teams. It connects CRM, customer interactions and Microsoft 365 work data to help sellers prepare, manage opportunities and follow up more efficiently. 
2. Is Microsoft Sales Agent the same as Microsoft 365 Copilot? 
   No. Microsoft 365 Copilot supports general productivity, while Sales Agent is focused on sales workflows such as customer insights, opportunity management, CRM updates and follow-ups. 
3. Does Sales Agent require a CRM system? 
   A CRM system is not the only data source, but it is usually important for achieving richer sales insights. Organizations with cleaner CRM data and clearer processes will benefit more. 
4. Is Sales Agent suitable for SMEs? 
   Yes, especially for SMEs with consultative selling, long sales cycles or relationship-driven accounts. It can help small sales teams handle more customer context with less administrative effort. 
5. How should organizations prepare for Sales Agent deployment? 
   They should first review CRM data quality, sales process consistency, customer ownership and Microsoft 365 usage maturity. A small pilot is recommended before full rollout. 

Microsoft Sales Agent is more than another AI feature. It helps sales teams bring customer knowledge, deal insights and follow-up actions into the flow of work. For Hong Kong businesses, the opportunity is to use AI not only to save time, but to improve sales execution, customer engagement and long-term account development. 

Want to explore how Microsoft Sales Agent can support your sales team?

Contact Superhub to assess your current sales workflow, identify practical AI use cases and plan your first Sales Agent pilot. 

Source: Sales Agent is now generally available: Bringing customer and deal intelligence into the flow of work | Microsoft Community Hub

Moving sales and service organizations forward with agentic CX and Microsoft 365 Copilot – Microsoft Dynamics 365 Blog

• Azure Content Understanding further supports AI Agents, RAG and enterprise knowledge management. 

• Key updates include GPT-5, Sync API, Semantic Chunking and Agentic Document Reasoning. 

• The priority for enterprises is not only to use a stronger model, but to organize documents into knowledge assets that AI can understand, search, and reference. 

• For organizations deploying Copilot, Enterprise Search, RAG, or AI Agents, document understanding is a key foundation for more accurate answers, less manual searching, and traceable enterprise knowledge. 

Many enterprises are already exploring Copilot, RAG and AI Agents. But in practice, the biggest blocker is often not the AI model — it is the knowledge foundation behind it. Critical information is still spread across PDFs, contracts, spreadsheets, scanned files, emails, audio recordings and video content. If AI cannot understand, search and reference this content properly, the result is often inaccurate answers, missing context and limited trust. 

The August update for Azure Content Understanding is a timely reminder that enterprise AI success depends on more than adopting the latest model. With GPT-5 model series support, Sync APIs, Semantic Chunking, Advanced Contextualization, and Agentic Document Reasoning, Azure Content Understanding helps organizations turn unstructured content into AI-ready knowledge assets that can support Copilot, Enterprise Search, RAG, and AI Agent scenarios. 

Image Source: From Sync APIs to support for the GPT-5 model series and agentic workflows: What’s new in Azure Content Understanding – August 2026 

The update can be viewed in two directions. CU 1.0 General Availability strengthens production scenarios with GPT-5 model support, lower token usage, improved grounding and confidence models. CU 2.0 Public Preview focuses on next-generation AI applications, introducing Sync API, Semantic Chunking, Advanced Contextualization and Agentic Document Reasoning to make it easier for enterprises to connect content into RAG, Enterprise Search and AI Agent workflows. 

A large portion of enterprise knowledge lives inside contracts, invoices, SOPs, customer records, technical documents and scanned files. If these assets are not organized, classified and structured, AI can return inaccurate search results, incorrect references or inconsistent answers. 

Azure Content Understanding helps bridge this gap by converting unstructured content into information that can be searched, indexed, referenced and reasoned over. For business leaders, this means a stronger foundation for AI adoption: fewer manual searches, more consistent answers, better traceability and clearer potential for productivity gains. 

1. GPT-5 Model Support: Balancing Accuracy, Cost and Speed

Enterprises can choose different GPT-5 model options based on the use case, balancing accuracy, speed and cost instead of applying high-cost models to every document workflow. 

2. Lower Token Cost and Better Accuracy: Making Large-Scale Document Processing More Practical

The new version improves grounding and confidence scoring to reduce token usage and increase output reliability, making contract review, form analysis and compliance document processing more cost-effective. 

3. Sync API: Supporting More Immediate Document Workflows

Sync API enables more immediate document processing workflows, suitable for customer onboarding, file submission, identity verification, insurance claims and application form pre-checking. This can shorten waiting time and reduce manual validation. 

4. Semantic Chunking: Improving RAG and Enterprise Search Quality

Semantic Chunking splits content based on document structure and meaning, helping preserve important context. This improves search and answer accuracy for RAG, Enterprise Search, and Copilot Extension scenarios. 

5. Agentic Document Reasoning: Handling More Complex Document Reasoning

Agentic Document Reasoning supports more complex document understanding and reasoning, making it suitable for high-risk scenarios such as contracts, financial reports, insurance documents and compliance files.

AI Agents cannot create enterprise knowledge out of nowhere. If documents are scattered, inconsistently named or stored only as PDFs, scanned files and email attachments, even a powerful model may struggle to provide stable, trustworthy and traceable answers. In the AI Agent era, the quality of data and knowledge management matters just as much as the model itself. 

From Superhub’s perspective, successful AI projects usually follow a practical foundation path: document understanding → knowledge base → RAG → AI Agent → workflow automation. This approach helps organizations move from scattered information to measurable AI use cases, instead of jumping directly into complex automation without a reliable data foundation. 

Document understanding is one layer of AI readiness. Before enterprises deploy AI at scale, they also need to understand whether their cloud environment, application landscape, data sources and operational processes can support the next step. 

Through recent Azure POV and cloud migration assessment work, Superhub has observed that many Hong Kong enterprises want to explore AI, Copilot or automation, but are still unclear about their current IT readiness, migration priorities and possible deployment roadmap. 

Azure POV provides a practical starting point by helping organizations assess existing infrastructure, applications, and operational processes, then identify whether they should begin with cloud migration, modernization, document understanding, RAG, Copilot Extension, or AI Agent scenarios. 

1. Identify high-value documents and data sources, such as contracts, SOPs, customer files, and technical support records. 
2. Build an AI-ready knowledge layer so content can be searched, indexed, referenced and traced. 
3. Start with high-value use cases such as knowledge search, compliance checking, contract summarization or document approval automation. 
4. If cloud and AI readiness are unclear, begin with an Azure POV assessment. 

Practical starting points include contract clause summarization, customer application pre-checking, internal SOP search, technical support knowledge base search and compliance document comparison. These use cases typically involve high document volume, repetitive work and high manual search effort, making productivity gains easier to measure. 

Before implementation, enterprises should also confirm three fundamentals: which information has the highest business value, whether access rights and compliance requirements are clear, and whether the cloud foundation can support search, indexing, reasoning and automation. These preparations help reduce AI project risk and support more accurate ROI estimation. 

The August update for Azure Content Understanding is more than a document processing upgrade. It reflects Microsoft’s direction in building the data foundation required for AI Agents, RAG and enterprise knowledge work. 

For Hong Kong enterprises, the real competitive advantage is not only using a more powerful AI model, but turning internal documents and knowledge into assets that AI can understand, search and apply. 

If your organization is planning to deploy Copilot, RAG, Enterprise Search, or AI Agents, the first step is to confirm whether your documents, cloud foundation, and data processes are AI-ready. Superhub can support this journey from free Azure POV assessment and cloud migration planning to Azure AI architecture design, helping identify the right starting point and turn internal knowledge into practical AI applications. 

1. What is Azure Content Understanding?

Azure Content Understanding is a capability in Microsoft Azure AI Foundry that helps enterprises extract and organize information from unstructured content such as documents, forms, images, audio, and video. It supports AI, RAG, enterprise search, and AI Agent applications. 

2. How is Azure Content Understanding different from traditional OCR?

OCR mainly converts images or scanned documents into text. Azure Content Understanding goes further by understanding document structure, meaning and context, making the content more suitable for AI use. 

3. Why do Copilot and AI Agents need document understanding?

The quality of Copilot or AI Agent responses depends on whether they can find accurate, complete and contextual enterprise information. Document understanding helps AI read and reference enterprise knowledge more accurately. 

4. If an enterprise wants to adopt Azure but is unsure where to start, what should it do?

Superhub provides a free Azure POV assessment to help organizations understand their current IT environment, cloud readiness, migration feasibility, cost considerations and possible Azure deployment direction. This gives enterprises a clearer starting point before moving into cloud modernization, AI adoption or automation projects. 

5. Should enterprises start with Azure POV or AI Agents first?

If cloud foundation, data governance and document management are not yet mature, it is better to start with Azure POV or an AI readiness assessment. AI Agents need a stable cloud and knowledge foundation, so assessing first usually leads to a stronger deployment path.

Source: Azure Content Understanding: Agentic workflow, Sync APIs, GPT-5 model series

• AI adoption is no longer just a productivity initiative or IT deployment project. It now involves identity security, data governance, AI risk monitoring, and compliance management. 

• For organizations adopting Microsoft 365 Copilot, AI agents, or custom AI workflows, the key challenge is often readiness of internal data, permissions, and governance. 

• To scale AI securely, businesses should strengthen three foundations first: identity security, data governance, and AI security controls. 

In the past, enterprise security mainly focused on devices, email, networks, and endpoint protection. However, as Copilot, AI agents, and enterprise knowledge search become more widely adopted, the security perimeter is shifting from the traditional network boundary to identity, data, and AI usage scenarios. 

Microsoft’s July 2026 Security update covers Project Perception, Microsoft Defender prompt injection protection preview, cloud agent posture and runtime protection, Microsoft Entra passkeys, and Microsoft Purview protection for AI and SaaS data risks. These updates show that Microsoft’s security direction is moving from protecting individual tools to protecting the entire AI operating environment. 

For business decision-makers, this means security strategy needs to be reconsidered. Organizations should not wait until Copilot or AI agents are already live before addressing risk. Identity, permissions, data classification, and AI usage policies should be handled during the AI project planning stage.

3. 1. AI Security Is Becoming a New Enterprise Security Battleground

   Generative AI can improve productivity, but it also introduces new risks such as prompt injection, data leakage, AI application misuse, and uncontrolled agent behavior. Microsoft’s latest AI security enhancements suggest that AI systems are becoming an important part of the enterprise security architecture. 

   What Changed 

   Microsoft Defender now includes prompt injection protection preview, designed to identify and isolate malicious AI instructions before they reach the mailbox. This helps reduce the risk of prompt injection attacks reaching users and AI assistants. At the same time, Microsoft is extending posture and runtime protection to agents built in Microsoft Foundry and Copilot Studio, as well as third-party managed agents, helping organizations manage AI-specific risk across their agent estate. 

   Business impact: Every Copilot, AI chatbot, or AI agent project should include security governance, permission design, and risk monitoring from the start. Businesses need to clearly define what data AI can access, what actions AI can perform, and whether approval, logging, and monitoring are required when AI workflows involve sensitive information. 

   2. Identity Is Becoming a Core Security Boundary

   In cloud and hybrid work environments, identity has become the most important security boundary. Most Microsoft 365, SaaS, and cloud services use user identity as the core of access control. If an account is compromised, attackers may be able to access email, Teams, SharePoint, OneDrive, and even internal systems without breaching a traditional network perimeter. 

   Microsoft Entra continues to advance passwordless authentication, passkeys, and multifactor authentication, with the goal of reducing account compromise, phishing, and social engineering risks. 

   Business impact: Strengthening identity security is not only about reducing account compromise. It also helps reduce business disruption, data leakage, and compliance risk. For organizations preparing to adopt Copilot, identity governance is the first step in AI readiness because Copilot accesses data based on the user’s identity and existing permissions. 

   3. Data Governance Is Critical to Copilot Success

   Copilot searches, summarizes, and generates content based on each user’s existing permissions. Therefore, the real risk is often not Copilot itself, but the organization’s existing data governance gaps, such as unclear SharePoint permissions, over-shared Teams content, unclassified sensitive data, and the lack of DLP or information protection label policies.

The Microsoft Purview updates further highlight the relationship between data security and AI governance. These capabilities help organizations identify sensitive data, apply information protection labels, manage DLP policies, and improve visibility into data movement across Microsoft 365, SaaS applications, and AI apps. For organizations preparing to adopt Copilot, these controls help IT teams and business leaders understand data risks before scaling AI usage more broadly.

Business Impact 

Strong data governance allows organizations to adopt Copilot and AI agents with greater confidence because sensitive data, over-shared content, and non-compliant access can be more easily identified and controlled. If SharePoint, Teams, and OneDrive permissions remain unmanaged, AI may amplify existing issues, turning productivity gains into higher data risk. 

Image Source: ​​​​What’s new in Microsoft Security: July 2026 

Developments such as Project Perception reflect a broader shift toward autonomous security. Newly announced by Microsoft, Project Perception is designed to coordinate specialized security agents that expose weaknesses, investigate cyberthreats, and help harden defenses. For Hong Kong organizations without large security teams, these AI-assisted capabilities may help reduce operational pressure and improve response speed over time.

• Security is moving from reactive response to AI-assisted detection and response. 

• Copilot adoption is no longer just a software deployment. It now involves security, governance, compliance, and change management. 

• The quality and security of AI outputs depend heavily on the organization’s data quality, permission design, and knowledge management maturity. 

In simple terms, security strategy in the AI era cannot rely on a single security product. Organizations need to govern identity, data, endpoints, cloud workloads, and AI agents together. For business leaders, the key is not simply to purchase another security tool, but to build a sustainable security operating model that enables AI adoption while maintaining data security and compliance. 

Based on Superhub’s experience helping Hong Kong organizations adopt Microsoft 365, Copilot, and AI solutions, the most common challenge is usually not the AI model itself, but data permissions, sensitive information management, and governance maturity. 

When Copilot or AI agents begin accessing enterprise knowledge bases, existing permission issues in SharePoint, Teams, and OneDrive can become more visible and more impactful. As a result, security strategy is shifting from network security toward identity, data, and AI governance. 

In practice, the most realistic approach is not to restructure all data at once. Organizations should start with high-risk scenarios, such as executive documents, customer data, financial information, HR files, external sharing sites, and SharePoint sites frequently used by Copilot pilot users. This allows businesses to establish an AI readiness baseline with lower risk and a shorter timeline before expanding to more departments and business processes. 

Organizations preparing to adopt Copilot, AI agents, or other AI-enabled workflows can start with the following actions:

1. Assess Microsoft Entra identity policies
   Review MFA, Conditional Access, passwordless authentication, and passkey deployment to reduce account compromise risk.
2. Review SharePoint, Teams, and OneDrive permissions
   Prioritize over-sharing, external sharing, and high-risk sites to prevent Copilot from amplifying existing data risks.
3. Establish data classification and sensitive information protection policies
   Use Microsoft Purview to define sensitivity labels, DLP policies, and data risk visibility.
4. Define governance for Copilot and AI agents
   Clearly define what data AI can access, approval processes, monitoring mechanisms, and user responsibilities. 
5. Start with pilot users and high-value scenarios
   Instead of addressing all data from day one, focus first on high-impact areas such as management, sales, finance, HR, or customer service. 

1. Does adopting Microsoft 365 Copilot increase data leakage risk?

Copilot does not bypass permissions, but it can expose existing over-sharing issues more easily. The key risk is unmanaged data access, not Copilot itself. 

2. What should organizations check before enabling Copilot?

Start with identity security, SharePoint and Teams permissions, external sharing, and sensitive data locations. Prioritize high-risk departments such as management, finance, HR, sales, and customer service. 

3. Is MFA enough, or should businesses consider passkeys?

MFA is still important, but passkeys further reduce phishing and password theft risk. Stronger identity protection is especially important when AI access depends on user permissions. 

4. Why is data governance important before AI adoption?

AI is only as safe and useful as the data it can access. Poor permissions, outdated content, or unclassified sensitive data can lead to wrong answers or higher data risk. 

5. How can Microsoft Purview support Copilot and AI readiness?

Purview helps identify sensitive data, apply protection labels, and manage DLP policies. It gives businesses better visibility before scaling Copilot or other AI use cases. 

The Microsoft Security July 2026 update shows that enterprise security is moving toward AI-assisted and agentic defense. AI security, identity security, and data governance are becoming key foundations for successful Copilot and AI agent deployment. For Hong Kong organizations, the next step is not to wait for AI technology to become more mature, but to establish a clear baseline for identity, permissions, data classification, and AI usage governance. Only when the security and governance foundation is clear can AI be deployed securely, compliantly, and sustainably while delivering measurable productivity gains and business value. 

Source: What’s new in Microsoft Security: July 2026 | Microsoft Security Blog

• Copilot adoption is not just a licensing project; it requires practical user enablement, business use cases and continuous guidance. 

• Organizations should help users apply Copilot to daily workflows such as meetings, email, documents, presentations, research and knowledge discovery. 

• Department use-case mapping, prompt guidance, training and agent starter kits help turn early interest into repeatable work habits. 

• Superhub supports the journey through Consult, Adopt and Review — from readiness alignment to practical enablement, adoption assets and expansion planning. 

• Security and governance remain important foundations for scalable adoption, but the main goal is to help users create measurable business value with Copilot. 

Microsoft 365 Copilot can improve productivity, but only when users understand where it fits into their daily work. Without the right adoption plan, organizations may assign licenses but still struggle to build consistent usage, department-level use cases and measurable business value. 

Superhub’s Copilot Readiness & Adoption Services help organizations move from license activation to practical adoption through executive alignment, user enablement, department use-case mapping, prompt and agent starter assets, and ongoing adoption review. Readiness and governance are included to support adoption at scale, not to replace the focus on business outcomes. 

To unlock the value of Copilot, organizations should focus on how people actually work. Users need practical examples, prompt guidance and department-specific scenarios that show how Copilot can support meetings, follow-ups, reporting, document drafting, presentations and knowledge search. 

Adoption becomes stronger when Copilot is connected to real workflows across departments such as sales, HR, finance, operations, marketing and management. Instead of treating Copilot as a general AI tool, each team should have clear scenarios that match their work priorities and expected outcomes. 

Readiness, security and governance still matter, especially before broader rollout. However, they should act as the foundation that enables confident adoption, while the main success measure remains whether users can apply Copilot consistently to create business value. 

Consult: Understand readiness and define the right starting point 

Align leadership on Copilot opportunities, business priorities and the right departments to start with. This helps define practical adoption goals before wider enablement begins. 

Adopt: Enable users to apply Copilot in daily work 

Through Copilot foundations training, prompt guidance, department use-case workshops and agent starter kits, we help users apply Copilot to practical work scenarios instead of stopping at basic feature exploration. 

Review: Measure impact and plan the next step 

Review usage, adoption, use case performance and future expansion opportunities to build a practical AI adoption roadmap for continuous improvement. 

This also includes reviewing new Copilot capabilities and identifying the next wave of use cases for business expansion, so organizations can continue improving adoption instead of treating Copilot as a one-off enablement activity. 

Superhub supports organizations across the Copilot adoption journey by combining business alignment, user enablement, practical adoption assets, readiness guidance and governance considerations based on each customer’s maturity and priorities. 

This approach helps organizations understand where to start, build user confidence, create department-specific scenarios, equip teams with reusable prompt and agent resources, and expand Copilot adoption in a controlled, measurable way.

Security and governance are still important when Copilot is expanded across more users and departments. Because Copilot respects existing Microsoft 365 permissions, organizations should understand data exposure, oversharing and sensitive information risks before scaling adoption. 

• Data visibility: Identify sensitive data and key exposure risks 

• Access and permissions: Review oversharing, broad access and stale permissions 

• Governance guidance: Recommend practical controls to support responsible AI adoption 

The goal is not to slow adoption down, but to help organizations scale Copilot with confidence. Governance provides guardrails, while training, use cases and adoption assets remain the main drivers of business value.

After the engagement, organizations gain practical guidance to support internal planning, stakeholder alignment and the next stage of Copilot adoption. This may include readiness observations, governance considerations, enablement recommendations and adoption planning insights, depending on the customer’s needs and priorities. 

These outputs help management teams understand key findings, opportunities, risks, recommended next steps and value-realization opportunities. They are designed to be shared internally and used to support the next stage of Copilot planning. 

A successful Copilot adoption program should help the organization move from experimentation to measurable business value. With the right training, use-case mapping, adoption assets, review process and governance foundation, organizations can expect: 

• Clearer adoption priorities across departments 

• Higher user confidence through practical enablement 

• More relevant use cases for daily work 

• Better visibility into readiness gaps, security risks and adoption progress 

• A repeatable AI adoption model that can grow with the business 

For business teams, this means users can start applying Copilot to real work faster, such as preparing meeting summaries, drafting documents, creating presentations, following up with customers or searching internal knowledge. 

For IT and management teams, it provides a clearer view of governance readiness, adoption progress and future expansion opportunities.

1. We already have Copilot licenses, but usage is low. What should we do?

Start with user enablement: Copilot training, use-case mapping, prompt guidance and department workshops. These help users move from testing Copilot features to applying Copilot in real work such as meetings, documents, email, presentations and collaboration. 

2. How can we identify the right Copilot use cases for each department?

Start by reviewing each department’s daily workflows, pain points and repetitive tasks. Superhub helps map Copilot scenarios to real business activities such as meeting preparation, proposal drafting, reporting, knowledge search, customer follow-up and process documentation. 

3. We are not sure which department or use case to start with. How can the package help?

They give users practical starting points instead of expecting them to create prompts or agent ideas from scratch. Department prompt packs and agent starter kits help users apply Copilot to familiar workflows and build confidence faster. 

4. How do prompt packs and agent starter kits help adoption?

Yes, but this should support adoption rather than delay it unnecessarily. Organizations should review Microsoft 365 readiness, licensing alignment, data exposure and permission oversharing so Copilot can be adopted with greater confidence. 

5. How can we make sure Copilot continues to deliver value?

Through regular adoption review and optimization planning, Superhub helps organizations review usage, business value, user feedback and future expansion opportunities. 

Deploying Copilot is not just about purchasing an AI tool. It is the starting point for helping users build new work habits with AI. To realize business value, organizations need practical training, use-case mapping, prompt and agent resources, adoption review, and a responsible governance foundation. 

If you are planning to deploy or expand Microsoft 365 Copilot, Superhub can help identify the right starting point, build user confidence, define high-value use cases and create a practical adoption roadmap for your organization.

• Microsoft MAI is not just a model announcement. It reflects Microsoft’s growing first-party enterprise AI capabilities. 

• Enterprise AI is moving from a single-model mindset to a multi-model strategy, where the focus is not the “best model” but the “right model for the right use case”. 

• As AI becomes part of business operations, governance, cost control, security, and access management become more important. 

• Azure AI Foundry is increasingly relevant as a platform for model deployment, governance, monitoring, and continuous management. 

• Hong Kong businesses should first clarify use cases, data readiness, risk requirements, and measurable outcomes before scaling AI adoption. 

Microsoft’s announcement of the MAI model family may look like another model update on the surface. But from an enterprise AI strategy perspective, it sends a clearer message: AI competition is shifting from “which single model is the most powerful” to “how multiple models can be used safely, effectively, and at scale.” 

For many Hong Kong businesses, AI is no longer just a tool to experiment with. It is starting to become part of daily operations. As this happens, the questions from leadership also change: Is this model suitable for our business workflow? Can the cost be controlled? Can data, access, and permissions be properly governed?

MAI, or Microsoft AI, refers to a family of AI models developed in-house by Microsoft AI. The new MAI model family covers capabilities such as reasoning, coding, voice, image generation and editing, and transcription. For enterprises, the key point is not simply that there are more models available. It shows Microsoft is building a more complete and more self-directed AI stack across model supply, platform integration, and enterprise deployment. 

Microsoft’s introduction of MAI should not be viewed only as a product update. The more important signal is that Microsoft is strengthening its control over core AI capabilities, including model availability, cost efficiency, performance optimization, and integration with platforms such as Azure and Copilot. 

For business leaders, this means AI adoption strategy can no longer stop at comparing model performance. The more practical questions are: Can this AI capability support long-term operations? Is the cost manageable? Does it meet the organisation’s security, compliance, and supplier risk requirements? 

What Is Multi-Model AI? 

The idea behind Multi-Model AI is simple: different tasks should use different models. Enterprises do not need to rely on one AI model for every workload. Instead, they can select the most suitable AI capability based on the use case, cost, speed, accuracy, and risk profile.

Why Is a Single Model No Longer Enough? 

• Document analysis may require stronger reasoning capability. 

• Customer service may require fast, real-time response. 

• Enterprise knowledge search may require secure data integration. 

• Content generation may require stronger creative capability. 

In other words, the enterprise question is changing from “which model is best?” to “which model is best suited to this business scenario?” This shift also makes AI governance, model routing, cost monitoring, and data permission management more important.

In the Hong Kong market, this shift is highly practical. Different organizations face different priorities. Some focus on compliance, some on customer service efficiency, while others want to improve internal document processing and knowledge management first. The value of Multi-Model AI is that businesses do not need to solve every problem with the same approach. 

• Financial services: Prioritize data sovereignty, compliance audit, and risk control to accelerate document review and internal knowledge search. 

• Retail and distribution: Focus on customer service automation, content generation, and operational efficiency to improve response speed while managing AI usage costs. 

• Manufacturing: Apply AI to document processing, knowledge management, and workflow automation to reduce repetitive manual work and retain internal know-how. 

What Is Azure AI Foundry? 

Azure AI Foundry is Microsoft’s enterprise AI platform for building, deploying, managing, and governing AI models, applications, and agents. As enterprises move beyond a single-model approach, the platform layer becomes more important because it helps connect models, security requirements, business applications, and governance controls into one operating environment. 

Platform Value Goes Beyond Model Availability 

For IT and business teams, the value of Azure AI Foundry is not only about which models are available. It is also about whether the organization can manage the following areas in a consistent way: 

• Model selection 

• Deployment management 

• Access control 

• Cost monitoring 

• Responsible AI management 

These capabilities directly affect whether an organization can move AI from isolated pilots into a monitored, governed, and sustainable operating environment.

AI Project Evaluation Framework 

When evaluating Microsoft Copilot, Azure AI, or other AI solutions, businesses should start with one simple question: does this project solve a real business problem? The following framework can help decision-makers assess whether an AI initiative is ready for practical adoption. 

Reminder for Decision-Makers 

Do not use model capability as the only evaluation criterion. 

The more important question is: 

Can this AI project create real business value in an environment that is governable, controllable, and scalable?

From our recent conversations with Hong Kong businesses, we are seeing a clear shift. Companies are no longer just asking, “What AI tools can we try?” They are asking, “How can we use AI safely and sustainably in daily work?” This shows that AI adoption is gradually moving from experimentation to operationalization. 

This step is often more complex than expected. Beyond choosing tools, organisations also need to address AI governance, Responsible AI, data permissions, user adoption, workflow integration, and long-term support. Many AI pilots fail to scale not because the model is not good enough, but because the organisation lacks a sustainable operating foundation. 

As a Hong Kong-based Microsoft Partner, AI Enablement company, and Managed Services Provider, Superhub helps organizations build a clearer and more practical enterprise AI roadmap — from Copilot and Azure AI readiness assessment, use case prioritization, data and access governance, to AI solution deployment and ongoing management.

The real meaning of Microsoft MAI is not only that Microsoft has added another family of AI models. It also reflects a broader shift towards the Multi-Model AI era. Future AI competition will not be determined by model capability alone. It will depend on whether enterprises can use AI in an environment that is governable, controllable, and scalable. 

For Hong Kong businesses, the next step is not necessarily to chase the latest model. It is to clarify business scenarios, data readiness, security requirements, and success metrics. Only then can AI move from a trial tool into a capability that delivers real productivity, efficiency, and business impact. 

If your organization is evaluating Microsoft Copilot, Azure AI, or an AI governance roadmap, Superhub can help you assess use cases, prepare data, strengthen security governance, and develop a clearer, executable AI adoption plan. 

1. What are Microsoft MAI models?

MAI models are a family of AI models developed in-house by Microsoft AI, covering areas such as reasoning, coding, voice, image generation and editing, and transcription. The key point is not only the models themselves, but also Microsoft’s broader strategy to build a more complete AI capability and platform ecosystem. 

2. Does MAI mean Microsoft will stop using OpenAI?

No. Public information does not indicate that Microsoft will stop working with OpenAI. A more practical interpretation is that Microsoft is expanding model choice, giving enterprises more options in terms of cost, flexibility, and long-term control. 

3. What is Multi-Model AI?

Multi-Model AI means selecting different AI models or capabilities for different business scenarios, such as document analysis, enterprise knowledge search, content generation, or customer service automation. 

4. What value does Azure AI Foundry bring to enterprises?

The value of Azure AI Foundry is not only model access. It helps enterprises build, deploy, monitor, and govern AI applications. When an organisation uses multiple models, unified management, security control, and cost monitoring become increasingly important. 

5. What should Hong Kong businesses assess before adopting AI?

Hong Kong businesses should first assess business use cases, data sensitivity, access management, compliance requirements, and expected outcomes. Rather than chasing the latest model, it is more important to confirm whether AI can safely solve a real business problem and support long-term execution.

Source: Building a hill-climbing machine: Launching seven new MAI models | Microsoft AI

Microsoft starts swapping OpenAI and Anthropic out for its own AI in some apps

• Microsoft has introduced OpenAI GPT-5.6 into Microsoft 365 Copilot and is rolling out Anthropic Claude Sonnet 5 to selected Copilot experiences. 

• The update expands Copilot’s ability to support content creation, analysis, planning, and multi-step work across familiar Microsoft 365 apps. 

• For business users, the real question is not whether GPT or Claude is stronger, but whether Copilot can help employees complete higher-value work faster. 

• Businesses should evaluate AI by productivity, adoption, workflow efficiency, governance, and measurable business outcomes. 

• For Hong Kong businesses, the key to AI success is not simply adopting the latest model, but embedding Copilot into real daily workflows with the right enablement and governance. 

According to Microsoft and OpenAI, GPT-5.6 is now available in Microsoft 365 Copilot across Word, Excel, PowerPoint, Copilot Chat, and Copilot Cowork. Claude Sonnet 5 is also rolling out to Copilot Cowork and PowerPoint. For business users, the key question is not “which model is strongest?”, but whether Copilot can help employees create documents, analyze data, build presentations, summarize meetings, and complete multi-step work more efficiently. This article explains what the update means for productivity, Copilot adoption, and AI return on investment. 

GPT-5.6 is now the preferred model in Microsoft 365 Copilot, supporting work across Word, Excel, PowerPoint, Copilot Chat, and Copilot Cowork. It is designed to help users move from rough ideas to more polished documents, clearer analysis, and stronger presentation drafts. 

Claude Sonnet 5 is also being introduced in Microsoft 365 Copilot, currently rolling out mainly to Copilot Cowork and PowerPoint. It is built for agentic, multi-step work such as planning, reasoning through context, and creating business artifacts. 

In short, Microsoft is not simply adding new model names. It is expanding the range of work Copilot can support inside the tools employees already use every day. 

Availability note
Claude Sonnet 5 should not be described as fully available across all Microsoft 365 Copilot apps. Based on Microsoft’s announcement, it is currently rolling out to Copilot Cowork and PowerPoint, and availability may vary by region and tenant configuration. 

Employees are already exposed to tools such as ChatGPT, Claude, Gemini, and Microsoft Copilot, so it is natural for business leaders to ask which AI model is best. In practice, there is rarely one simple answer. Different models may perform better across writing, reasoning, summarization, or multi-step work. The business priority should be clear: does the AI capability improve the way employees work?

Different teams need AI for different reasons. Marketing teams need to draft articles, proposals, EDMs, and social posts faster. Management teams need to compare options and analyze information. Project teams need help with plans, task organization, and knowledge sharing. The value of Copilot is not limited to chat; it comes from supporting real work across departments. 

The real value of this update is not the model upgrade itself, but the wider range of business tasks Copilot can now support. 

For example, GPT-5.6 can help users draft documents, analyze data, create presentations, and work through open-ended problems in Microsoft 365 apps. Claude Sonnet 5 adds another option for multi-step planning and business content creation in Copilot Cowork and PowerPoint. 

For most businesses, the practical questions are simple: 

• Can employees complete work faster? 

• Can the quality of work improve? 

• Can repetitive work be reduced? 

These are the questions that define real business value.

For knowledge workers, productivity gains usually come from faster document creation, better information organization, and less time spent preparing materials. However, these benefits only appear when employees know how to apply Copilot to their daily work. 

From Superhub’s experience working with business customers, many companies have already deployed or purchased Microsoft 365 Copilot, but employees may not be using its full capabilities. Some still treat Copilot mainly as a general chat tool, instead of applying it to meeting summaries, document drafting, Excel analysis, presentation creation, or cross-department knowledge search. As a result, adoption remains low, productivity gains are limited, and the return on AI investment becomes harder to justify. 

That is why businesses should not evaluate GPT-5.6, Claude Sonnet 5, or any new AI capability in isolation. A more effective approach is to identify high-frequency, repetitive, and time-consuming workflows first, then improve adoption through training, departmental pilots, and usage tracking. 

As AI becomes more capable, businesses should review how it is managed. Key areas include: 

• Data access permissions 

• Protection of company data 

• AI usage policies 

• Compliance requirements 

• Employee usage boundaries 

Businesses can start with three steps: identify repetitive or information-heavy workflows, define clear AI usage standards, and validate results through departmental pilots. For organizations that have already purchased Copilot, the priority is not to wait for the next AI feature. It is to use enablement, use case design, and adoption tracking to help employees apply Copilot in daily work. 

GPT-5.6 becoming part of Microsoft 365 Copilot and Claude Sonnet 5 rolling out to selected Copilot experiences are both worth attention. But for most business users, the real focus is whether Copilot can help employees complete daily work faster, reduce repetitive tasks, and improve the quality of content, analysis, and collaboration. For organizations that have already purchased Copilot, the next step is to review whether employees are using it well, whether the right use cases and training are in place, and whether governance is strong enough to turn AI licensing cost into practical productivity gains and ROI. 

If your organization has already deployed Microsoft 365 Copilot but adoption remains low, Superhub can help review current usage, identify practical use cases, and design enablement plans to improve adoption and ROI. 

1. What does GPT-5.6 mean for Microsoft 365 Copilot users?
   It gives Copilot stronger support for content creation, analysis, and knowledge work across Microsoft 365 apps. For business users, the value is faster, higher-quality work in documents, presentations, data analysis, and decision preparation. 
2. How does Claude Sonnet 5 improve Copilot workflows?
   Claude Sonnet 5 can support more complex workflows such as multi-step planning, content organization, and business content creation. Based on Microsoft’s announcement, it is currently rolling out to Copilot Cowork and PowerPoint, not all Microsoft 365 Copilot apps. 
3. Should businesses directly compare GPT-5.6 and Claude Sonnet 5?
   Not as a standalone decision. Businesses should evaluate AI by use case, productivity impact, output quality, governance fit, and adoption potential. 
4. Can AI model choice affect employee productivity?
   Yes, but productivity does not come from the model alone. It depends on whether AI connects with company data, fits daily tools, and supports real workflows such as meeting preparation, reporting, and data analysis. 
5. How is Microsoft 365 Copilot different from standalone AI tools?
   Microsoft 365 Copilot works inside Word, Excel, PowerPoint, Outlook, Teams, and Microsoft Graph, so AI can operate within the company’s existing work environment instead of becoming another separate tool. 

Source: Available today: OpenAI’s GPT-5.6 in Microsoft 365 Copilot | Microsoft Community Hub

Available today: Anthropic’s Claude Sonnet 5 in Microsoft 365 Copilot | Microsoft Community Hub

Nan Fung Property Management (NFPM), with a workforce of approximately 1,800 employees, is a premier property management company in Hong Kong providing comprehensive services for more than 80 projects including luxury residences, Grade A commercial buildings, shopping malls, and industrial buildings.

Guided by its vision, “We Improve the Quality of Life”, NFPM is committed to delivering reliable, high‑quality property and facility management services. From an IT and operational leadership perspective, NFPM continues to prioritize service reliability, operational excellence, and technology‑enabled improvements to support its business objectives.

Nan Fung Property Management required a secure and reliable environment to enable the consistent use of Generative AI, while ensuring productivity gains without compromising data security or staff experience.

Key challenges identified include:

• Lack of a Secure and Centralized AI Platform

Lack of a secure and centralized Generative AI platform resulted in fragmented use of public and free tools, creating security, compliance, and workflow consistency risks.

• Inefficient and Error‑Prone Information Access

Manual information searches were time‑consuming and error‑prone, making it difficult for employees to quickly locate operational procedures and guidelines, which negatively impacted productivity and staff experience.

Partnering with SUPERHUB, Nan Fung Property Management developed its proprietary Generative AI application, NFPM A.I., to establish a secure platform that enables staff to safely leverage Generative AI at work while protecting data and information.

Key solution highlights include:

• Generative AI platform:

NFPM A.I. provides a secure and centralized environment for staff to access advanced AI models in alignment with security and data protection requirements.

• Intelligent AI chatbot:

NFPM A.I. enables intuitive interaction with Generative AI in a protected environment, helping teams work more efficiently and confidently.

• RAG-powered knowledge access – VIVA Genie:

VIVA Genie allows staff to retrieve NFPM‑specific policies, procedures, and guidelines through natural conversation, ensuring accurate and context‑relevant responses.

• Improved productivity and reduced risk:

The solution minimizes reliance on high‑risk public tools, delivers instant and reliable information, and enables teams to focus on higher‑value, mission‑critical tasks.

Nan Fung Property Management selected SUPERHUB for its expertise in AI solutions and its ability to deliver a secure environment for business data. Throughout the project, SUPERHUB worked closely with NFPM to understand requirements and address technical challenges effectively.

• AI expertise delivering secure and trusted solutions:

Delivered AI solutions in a safeguarded environment, ensuring data and information were protected.

• Proactive and responsive delivery:

Maintained close collaboration with NFPM, promptly addressing requirements and resolving technical issues throughout the project.

• Positive operational and staff impact:

The collaboration supported improvements in operational efficiency and staff experience through the successful deployment of Generative AI capabilities.

Staff experience and productivity at Nan Fung Property Management have improved, as teams moved beyond basic information retrieval to integrating Generative AI into their daily work. By supporting more complex productivity tasks, the platform enables employees to focus on higher‑impact, value‑adding responsibilities.

Key results achieved include:

• Improved staff productivity and work experience: Generative AI is embedded into day‑to‑day workflows, enabling teams to handle more complex tasks efficiently while focusing on higher‑value work.
• AI as a practical co‑pilot: The solution supports staff with direct, consolidated responses and task assistance through natural language interaction.
• Time and effort savings: By aggregating intelligence from multiple internal sources within a secure environment, the platform reduces time spent searching for information and supports more efficient day‑to‑day operations.

Broader Impact

This success story highlights how Generative AI can be applied in practical and secure ways to support everyday operations. It also demonstrates how thoughtful AI adoption can create meaningful benefits for both the business and its people.

Looking ahead, Nan Fung Property Management plans to further optimize the AI platform to support additional operational areas and business functions, expanding its capabilities to address a broader range of complex queries and tasks and enabling wider adoption across NFPM.

This ongoing development reinforces NFPM’s commitment to continuous improvement and operational excellence. By further leveraging generative AI across projects, NFPM aims to support business growth while ensuring its technology foundation remains scalable, resilient, and aligned with long‑term strategic objectives – advancing NFPM’s vision, “We Improve the Quality of Life”.

As businesses adopt Microsoft 365, Azure, generative AI, and hybrid work, cloud security is becoming more complex. Many organizations assume that moving data to the cloud means it is automatically protected. In reality, data access, identity management, AI governance, and recovery readiness remain business responsibilities. 

This article highlights five cloud security gaps businesses often overlook, and how Hong Kong organizations can build a stronger foundation for secure AI and cloud adoption. 

Many business leaders believe that storing data in Microsoft 365, Azure, or other major cloud platforms means their security concerns have already been handled. 

Cloud providers secure the underlying infrastructure, but businesses remain responsible for protecting data, managing access, securing identities, and planning recovery. This is known as the Shared Responsibility Model. 

Even when the platform is secure, misconfiguration, excessive permissions, and weak governance can still expose business data. 

Leadership Consideration:
Cloud adoption should be treated as an ongoing governance discipline, not a one-off deployment project. 

Hybrid work, SaaS applications, third-party integrations, and generative AI have expanded how business data is accessed and shared. 

Microsoft 365 Copilot works with business content a user already has permission to access. If permissions and data governance are not well managed, AI can amplify existing visibility and access issues. 

For many Hong Kong businesses, SharePoint, Teams, OneDrive, and SaaS tools are now core to daily operations, but permission reviews, external sharing controls, and data classification often lag behind. 

Business Impact:
Traditional network security alone is not enough for AI-driven collaboration and modern data flows. 

Gap 1: Identity Security 

Identity security is now the foundation of modern cybersecurity. If an attacker gains access to an employee account, they may reach sensitive data without breaching a traditional firewall. 

Business Risk:
Account compromise, privilege escalation, and lateral movement. 

Gap 2: Over-Sharing and Weak Access Control 

SharePoint sites or Teams workspaces created years ago may never have been reviewed, leaving employees with access to documents that no longer match their role. 

Business Risk: Sensitive documents may be accessed or shared by the wrong users. 

Operational Impact: Loss of information control and exposure of business-sensitive data. 

Gap 3: AI and Data Governance Risk 

AI works based on existing access controls. If data is poorly classified or permissions are inconsistent, AI tools may surface sensitive information in unintended ways. 

Business Risk: Confidential information may be surfaced in unintended contexts. 

Operational Impact: Reduced trust in AI adoption and lower confidence in business data. 

Gap 4: Compliance and Regulatory Exposure 

If a business cannot demonstrate how data is accessed, retained, protected, and governed, regulatory and audit exposure can increase. 

Business Risk: Audit gaps, regulatory pressure, and compliance failure. 

Operational Impact: Additional legal cost, delayed audits, and reputational damage. 

Gap 5: Backup and Recovery Readiness 

Even with strong protection, accidental deletion, ransomware, system failure, or data corruption can still happen. Businesses need recovery readiness, not just prevention. 

Business Risk: Inability to restore critical business data quickly. 

Operational Impact: Downtime, revenue loss, and customer disruption. 

How SUPERHUB Adds Value:
As a Hong Kong-based Microsoft Solutions Partner and MSP, SUPERHUB helps businesses identify practical cloud security gaps across Microsoft 365 permissions, Entra ID, Defender, Purview, and backup readiness — turning security tools into sustainable daily governance. 

u003cbu003eu003cspan data-contrast=u0022autou0022u003eThe Evolution of Microsoft Teamsu003c/spanu003eu003c/bu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003ernrnu003cspan data-contrast=u0022autou0022u003eIn the past, Teams was mainly used for three core purposes:u003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003ernu003culu003ern tu003cli aria-setsize=u0022-1u0022 data-leveltext=u0022u0022 data-font=u0022Symbolu0022 data-listid=u00222u0022 data-list-defn-props=u0022{u0026quot;335552541u0026quot;:1,u0026quot;335559685u0026quot;:720,u0026quot;335559991u0026quot;:360,u0026quot;469769226u0026quot;:u0026quot;Symbolu0026quot;,u0026quot;469769242u0026quot;:[8226],u0026quot;469777803u0026quot;:u0026quot;leftu0026quot;,u0026quot;469777804u0026quot;:u0026quot;u0026quot;,u0026quot;469777815u0026quot;:u0026quot;multilevelu0026quot;}u0022 data-aria-posinset=u00221u0022 data-aria-level=u00221u0022u003eu003cspan data-contrast=u0022autou0022u003eTeam chatu003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003eu003c/liu003ernu003c/ulu003ernu003culu003ern tu003cli aria-setsize=u0022-1u0022 data-leveltext=u0022u0022 data-font=u0022Symbolu0022 data-listid=u00222u0022 data-list-defn-props=u0022{u0026quot;335552541u0026quot;:1,u0026quot;335559685u0026quot;:720,u0026quot;335559991u0026quot;:360,u0026quot;469769226u0026quot;:u0026quot;Symbolu0026quot;,u0026quot;469769242u0026quot;:[8226],u0026quot;469777803u0026quot;:u0026quot;leftu0026quot;,u0026quot;469777804u0026quot;:u0026quot;u0026quot;,u0026quot;469777815u0026quot;:u0026quot;multilevelu0026quot;}u0022 data-aria-posinset=u00222u0022 data-aria-level=u00221u0022u003eu003cspan data-contrast=u0022autou0022u003eVideo meetingsu003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003eu003c/liu003ernu003c/ulu003ernu003culu003ern tu003cli aria-setsize=u0022-1u0022 data-leveltext=u0022u0022 data-font=u0022Symbolu0022 data-listid=u00222u0022 data-list-defn-props=u0022{u0026quot;335552541u0026quot;:1,u0026quot;335559685u0026quot;:720,u0026quot;335559991u0026quot;:360,u0026quot;469769226u0026quot;:u0026quot;Symbolu0026quot;,u0026quot;469769242u0026quot;:[8226],u0026quot;469777803u0026quot;:u0026quot;leftu0026quot;,u0026quot;469777804u0026quot;:u0026quot;u0026quot;,u0026quot;469777815u0026quot;:u0026quot;multilevelu0026quot;}u0022 data-aria-posinset=u00223u0022 data-aria-level=u00221u0022u003eu003cspan data-contrast=u0022autou0022u003eDocument collaborationu003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003eu003c/liu003ernu003c/ulu003ernu003cspan data-contrast=u0022autou0022u003eBased on the product direction we have seen over the past two years, the role of Teams is clearly changing.u003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003ernrnu003cbu003eu003cspan data-contrast=u0022autou0022u003eTraditionalu003c/spanu003eu003c/bu003eu003cbu003eu003cspan data-contrast=u0022autou0022u003e Teamsu003c/spanu003eu003c/bu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003ernu003culu003ern tu003cli aria-setsize=u0022-1u0022 data-leveltext=u0022u0022 data-font=u0022Symbolu0022 data-listid=u00223u0022 data-list-defn-props=u0022{u0026quot;335552541u0026quot;:1,u0026quot;335559685u0026quot;:720,u0026quot;335559991u0026quot;:360,u0026quot;469769226u0026quot;:u0026quot;Symbolu0026quot;,u0026quot;469769242u0026quot;:[8226],u0026quot;469777803u0026quot;:u0026quot;leftu0026quot;,u0026quot;469777804u0026quot;:u0026quot;u0026quot;,u0026quot;469777815u0026quot;:u0026quot;multilevelu0026quot;}u0022 data-aria-posinset=u00221u0022 data-aria-level=u00221u0022u003eu003cspan data-contrast=u0022autou0022u003eCentered on human-to-human communicationu003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003eu003c/liu003ernu003c/ulu003ernu003culu003ern tu003cli aria-setsize=u0022-1u0022 data-leveltext=u0022u0022 data-font=u0022Symbolu0022 data-listid=u00223u0022 data-list-defn-props=u0022{u0026quot;335552541u0026quot;:1,u0026quot;335559685u0026quot;:720,u0026quot;335559991u0026quot;:360,u0026quot;469769226u0026quot;:u0026quot;Symbolu0026quot;,u0026quot;469769242u0026quot;:[8226],u0026quot;469777803u0026quot;:u0026quot;leftu0026quot;,u0026quot;469777804u0026quot;:u0026quot;u0026quot;,u0026quot;469777815u0026quot;:u0026quot;multilevelu0026quot;}u0022 data-aria-posinset=u00222u0022 data-aria-level=u00221u0022u003eu003cspan data-contrast=u0022autou0022u003eUsers manually search for informationu003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003eu003c/liu003ernu003c/ulu003ernu003culu003ern tu003cli aria-setsize=u0022-1u0022 data-leveltext=u0022u0022 data-font=u0022Symbolu0022 data-listid=u00223u0022 data-list-defn-props=u0022{u0026quot;335552541u0026quot;:1,u0026quot;335559685u0026quot;:720,u0026quot;335559991u0026quot;:360,u0026quot;469769226u0026quot;:u0026quot;Symbolu0026quot;,u0026quot;469769242u0026quot;:[8226],u0026quot;469777803u0026quot;:u0026quot;leftu0026quot;,u0026quot;469777804u0026quot;:u0026quot;u0026quot;,u0026quot;469777815u0026quot;:u0026quot;multilevelu0026quot;}u0022 data-aria-posinset=u00223u0022 data-aria-level=u00221u0022u003eu003cspan data-contrast=u0022autou0022u003eWorkflows are mainly executed by peopleu003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003eu003c/liu003ernu003c/ulu003ernu003cbu003eu003cspan data-contrast=u0022autou0022u003eNext-Generation Teamsu003c/spanu003eu003c/bu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003ernu003culu003ern tu003cli aria-setsize=u0022-1u0022 data-leveltext=u0022u0022 data-font=u0022Symbolu0022 data-listid=u00224u0022 data-list-defn-props=u0022{u0026quot;335552541u0026quot;:1,u0026quot;335559685u0026quot;:720,u0026quot;335559991u0026quot;:360,u0026quot;469769226u0026quot;:u0026quot;Symbolu0026quot;,u0026quot;469769242u0026quot;:[8226],u0026quot;469777803u0026quot;:u0026quot;leftu0026quot;,u0026quot;469777804u0026quot;:u0026quot;u0026quot;,u0026quot;469777815u0026quot;:u0026quot;multilevelu0026quot;}u0022 data-aria-posinset=u00221u0022 data-aria-level=u00221u0022u003eu003cspan data-contrast=u0022autou0022u003eAI helps handle calls and routine enquiriesu003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003eu003c/liu003ernu003c/ulu003ernu003culu003ern tu003cli aria-setsize=u0022-1u0022 data-leveltext=u0022u0022 data-font=u0022Symbolu0022 data-listid=u00224u0022 data-list-defn-props=u0022{u0026quot;335552541u0026quot;:1,u0026quot;335559685u0026quot;:720,u0026quot;335559991u0026quot;:360,u0026quot;469769226u0026quot;:u0026quot;Symbolu0026quot;,u0026quot;469769242u0026quot;:[8226],u0026quot;469777803u0026quot;:u0026quot;leftu0026quot;,u0026quot;469777804u0026quot;:u0026quot;u0026quot;,u0026quot;469777815u0026quot;:u0026quot;multilevelu0026quot;}u0022 data-aria-posinset=u00222u0022 data-aria-level=u00221u0022u003eu003cspan data-contrast=u0022autou0022u003eAI summarizes meetings and business contextu003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003eu003c/liu003ernu003c/ulu003ernu003culu003ern tu003cli aria-setsize=u0022-1u0022 data-leveltext=u0022u0022 data-font=u0022Symbolu0022 data-listid=u00224u0022 data-list-defn-props=u0022{u0026quot;335552541u0026quot;:1,u0026quot;335559685u0026quot;:720,u0026quot;335559991u0026quot;:360,u0026quot;469769226u0026quot;:u0026quot;Symbolu0026quot;,u0026quot;469769242u0026quot;:[8226],u0026quot;469777803u0026quot;:u0026quot;leftu0026quot;,u0026quot;469777804u0026quot;:u0026quot;u0026quot;,u0026quot;469777815u0026quot;:u0026quot;multilevelu0026quot;}u0022 data-aria-posinset=u00223u0022 data-aria-level=u00221u0022u003eu003cspan data-contrast=u0022autou0022u003eAI helps employees find knowledge and information fasteru003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003eu003c/liu003ernu003c/ulu003ernu003culu003ern tu003cli aria-setsize=u0022-1u0022 data-leveltext=u0022u0022 data-font=u0022Symbolu0022 data-listid=u00224u0022 data-list-defn-props=u0022{u0026quot;335552541u0026quot;:1,u0026quot;335559685u0026quot;:720,u0026quot;335559991u0026quot;:360,u0026quot;469769226u0026quot;:u0026quot;Symbolu0026quot;,u0026quot;469769242u0026quot;:[8226],u0026quot;469777803u0026quot;:u0026quot;leftu0026quot;,u0026quot;469777804u0026quot;:u0026quot;u0026quot;,u0026quot;469777815u0026quot;:u0026quot;multilevelu0026quot;}u0022 data-aria-posinset=u00224u0022 data-aria-level=u00221u0022u003eu003cspan data-contrast=u0022autou0022u003eAI participates in selected business workflowsu003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003eu003c/liu003ernu003c/ulu003ernu003cspan data-contrast=u0022autou0022u003eThis reflects Microsoft’s direction of embedding AI into the full work experience, rather than limiting it to a chat window or a standalone Copilot interface.u003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003ernrnu003cspan data-contrast=u0022autou0022u003eIn other words:u003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003ernrnu003cspan data-contrast=u0022autou0022u003eEmployees will not need to intentionally “open AI” to use it.u003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003ernrnu003cspan data-contrast=u0022autou0022u003eAI will appear naturally as work happens.u003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003ernrnu003cbu003eu003cspan data-contrast=u0022autou0022u003eDecision Maker Noteu003c/spanu003eu003c/bu003ernu003cspan data-contrast=u0022autou0022u003eThis model can significantly lower the barrier to AI adoption and help organizations increase AI usage across daily work.u003c/spanu003eu003cspan data-ccp-props=u0022{}u0022u003e u003c/spanu003e

When these gaps are ignored, the impact extends beyond IT. Operations, finance, and customer service teams may all face disruption. 

For Hong Kong businesses, the most serious impacts often include: 

• Reputational damage 

• Reduced customer trust 

• Higher compliance exposure 

• Longer business downtime 

• Increased operational cost 

Decision Maker Note:
The true cost of a security incident includes downtime, recovery effort, lost productivity, and reputational damage. 

From SUPERHUB’s perspective, the challenge is no longer just stopping attackers at the firewall. It is ensuring AI tools only access appropriate data, and employees only have the permissions they need. 

As Microsoft 365 Copilot, AI agents, and automation scale, identity security and data governance are becoming board-level priorities.

To build an AI-ready security foundation, businesses should start with identity-first security, treating authentication and access control as the new frontline of protection. 

• An AI governance framework 

• A data classification policy 

• A security monitoring process 

• An incident response plan 

• A backup validation process 

• A recovery testing schedule 

Microsoft’s modern security architecture supports Zero Trust principles across identity, threat protection, data governance, and compliance. 

Microsoft Entra ID helps strengthen authentication, conditional access, and secure user access. 

Microsoft Defender helps detect, investigate, and respond to threats across users, devices, applications, and cloud workloads. 

Microsoft Purview supports data classification, information protection, compliance, and governance. 

How SUPERHUB Helps:
SUPERHUB supports assessment, planning, deployment, and ongoing management — including Microsoft 365 security reviews, Copilot readiness assessments, permission checks, data classification, security monitoring, and managed services. 

• Cloud security is not the same as data security. 

• Identity security and data governance are becoming more important in the AI era. 

• Over-permissioned access and uncontrolled data sharing are major hidden risks. 

• Backup and recovery readiness remain essential to business resilience. 

• Security, governance, and AI adoption should be treated as one connected transformation agenda. 

1. Does storing data in Microsoft 365 mean it is fully protected automatically?
   No. Microsoft secures the platform, but businesses still manage access, identity, governance, and recovery.
2. What is the Shared Responsibility Model?
   It means the cloud provider secures the infrastructure, while the business remains responsible for data, accounts, access control, and governance.
3. Why is identity security important in cloud environments?
   Most cloud access depends on identity. A compromised account can expose business data without breaching a network boundary.
4. Does AI create new data security risks?
   Yes. AI can amplify existing permission and governance issues, so access control and data classification are essential.
5. Why do cloud environments still need backup and recovery planning?
   Because accidental deletion, ransomware, operational errors, and service disruption can still happen. 

Moving to the cloud does not automatically make a business secure. In the AI era, identity management, data governance, and recovery readiness are essential to protecting business operations. 

For Hong Kong businesses, the goal is not only to defend against threats, but to enable secure cloud and AI adoption while maintaining trust, compliance, and continuity. 

If your organization is preparing for Microsoft 365 Copilot or strengthening Microsoft 365 and Azure security governance, SUPERHUB can help assess identity permissions, data classification, external sharing, Microsoft Purview configuration, backup readiness, and ongoing managed services.

Source: Best Practices for Cloud Data Security

The Microsoft Teams June 2026 update introduces several new capabilities across AI, voice agents, intelligent meetings and Teams Rooms. But beyond the feature list, the bigger message is clear: Microsoft Teams is evolving from a meeting and communication tool into an AI-powered work platform. 

For Hong Kong businesses, the key question is no longer whether to use AI. It is whether the organization is ready to let AI participate safely and effectively in daily work. From AI agents and automated workflows to new forms of human-AI collaboration, business leaders need to rethink process design, knowledge management and governance. 

The next competitive advantage will not come from deploying more AI tools. It will come from building a secure, scalable and supported operating model where people, data, workflows and AI can work together with confidence. 

• Microsoft Teams is evolving from a collaboration tool into an AI-powered work platform. 

• AI is moving beyond task assistance and starting to participate in workflows and operations. 

• The direction of Teams aligns closely with Microsoft’s broader agentic AI strategy. 

• Future AI value will be measured less by individual productivity and more by process redesign. 

• Hong Kong businesses should prepare their governance, workflows and data foundation before scaling AI adoption. 

Key Takeaway
The enterprise AI journey is entering its next stage. The focus is no longer simply on deploying AI, but on making AI a manageable, governed and scalable work partner. 

Microsoft’s recent Microsoft Teams June 2026 – InfoComm Edition update introduces new capabilities across AI, voice agents, intelligent meetings and Teams Rooms. 

However, if these updates are viewed only as a product feature release, businesses may miss the bigger signal behind them. 

Across Microsoft Build 2026, the Work Trend Index 2026 and the expanding Microsoft 365 Copilot ecosystem, Microsoft is pointing toward one clear direction: 

AI is shifting from a productivity tool into a digital workforce that can participate in business workflows. 

Microsoft Teams is becoming a central layer in this transformation. 

It is no longer just a place for meetings and chats. It is becoming the entry point where company knowledge, team collaboration, business processes and AI agents come together. 

Business Impact
Organizations may soon use Teams less like a traditional meeting system and more like a daily work operating platform. 

Source: Microsoft Teams Blog – What’s New in Microsoft Teams | June 2026 – InfoComm Edition 

The Evolution of Microsoft Teams 

In the past, Teams was mainly used for three core purposes: 

• Team chat 

• Video meetings 

• Document collaboration 

Based on the product direction we have seen over the past two years, the role of Teams is clearly changing. 

Traditional Teams 

• Centered on human-to-human communication 

• Users manually search for information 

• Workflows are mainly executed by people 

Next-Generation Teams 

• AI helps handle calls and routine enquiries 

• AI summarizes meetings and business context 

• AI helps employees find knowledge and information faster 

• AI participates in selected business workflows 

This reflects Microsoft’s direction of embedding AI into the full work experience, rather than limiting it to a chat window or a standalone Copilot interface. 

In other words: 

Employees will not need to intentionally “open AI” to use it. 

AI will appear naturally as work happens. 

Decision Maker Note
This model can significantly lower the barrier to AI adoption and help organizations increase AI usage across daily work. 

Beyond Productivity: Redesigning How Work Gets Done 

In the past, most business discussions around AI focused on improving individual productivity. 

The next stage of value, however, will likely come from redesigning how work gets done. 

People 

AI can begin to take on repetitive and administrative work. 

For example: 

• Meeting coordination 

• Enquiry handling 

• Information consolidation 

• Basic customer service support 

This allows employees to spend more time on judgement, coordination and decision-making. 

Knowledge 

Source: Microsoft Teams Blog – What’s New in Microsoft Teams | June 2026 – InfoComm Edition 

For many organizations, the challenge is not a lack of information. 

The bigger challenge is that information is often scattered across systems, teams and documents. 

Microsoft Graph acts as the data and relationship layer across Microsoft 365, connecting emails, files, meetings and knowledge to provide enterprise context for Microsoft 365 Copilot. 

When AI can better understand organizational knowledge, the cost of finding information can be significantly reduced. 

Operations 

AI agents are starting to enter business workflows. 

This means AI is no longer just answering questions. It is starting to help move work forward. 

For example: 

• Customer service routing 

• Internal knowledge lookup 

• Task follow-up 

• Workflow notifications 

Business Impact 

Future business value will increasingly come from: 

• Faster decision-making 

• Less process friction 

• More efficient knowledge flow 

not simply from saving a few minutes of work. 

From Copilot to Agentic Work 

When we look at the Teams update within Microsoft’s broader AI roadmap, the direction becomes even clearer. 

Phase 1: AI Assistants 

Representative product: 

• Microsoft 365 Copilot 

AI mainly helps with: 

• Content creation 

• Question answering 

• Thinking and ideation support 

Phase 2: Human-AI Collaboration 

Representative products: 

• Copilot Co-work 

• AI collaboration tools 

AI starts to work alongside employees to complete tasks. 

Phase 3: Agentic AI 

Source: Microsoft Teams Blog – What’s New in Microsoft Teams | June 2026 – InfoComm Edition 

Representative trends: 

• AI Agents 

• Teams Phone Agent 

• Microsoft Scout 

• Copilot Studio Agents 

AI begins to execute parts of business workflows on behalf of the organization. 

This aligns closely with the “Frontier Firms” concept introduced in Microsoft’s Work Trend Index 2026. Frontier Firms are organizations that redesign how people and AI work together, and embed AI into their operating model.

Many organizations in financial services, legal services, property management and logistics are facing common pressures: 

• Shortage of skilled talent 

• Increasing compliance requirements 

• Higher expectations for customer response speed 

AI agents can become a new operating lever for these businesses. 

But the value is only sustainable when AI is deployed within a governed and monitored framework. 

Leadership Consideration 

The future competition may not be about “who has AI”. 

It will be about who can manage an AI-enabled workforce effectively. 

As Teams, Copilot and AI agents become part of daily work, business leaders should assess four foundational readiness areas before deploying more standalone AI tools. 

• Are core workflows documented, repeatable and ready for automation? 

• Is company knowledge structured across Teams, SharePoint and Microsoft 365? 

• Are access rights, identity controls and data governance clearly defined? 

• Can IT teams monitor, secure and scale AI usage across departments? 

If the answer is unclear, AI adoption may create more complexity than value. A readiness-first approach helps businesses turn AI from an experimental tool into a scalable operating capability. 

Teams Is Becoming an AI Operating Layer 

From Superhub’s perspective, the biggest significance of this Teams update is not the number of new features. 

What matters more is this: 

Microsoft is embedding AI directly into the environments where employees already work every day. 

This means businesses may not need to move into a separate AI platform to work with AI. 

Instead, AI will increasingly exist inside Teams, Outlook, SharePoint and daily business workflows. 

This is consistent with the direction we have seen across Build 2026, Work Trend Index 2026 and the broader Agentic AI trend. 

Many Hong Kong businesses are still at the stage of: 

• Individual employees using AI on their own 

• Department-level pilots 

• Standalone productivity tools 

Frontier Firms, however, are already asking deeper questions: 

• Which workflows can be supported by AI? 

• How should responsibilities between people and AI be defined? 

• How can AI be governed and monitored continuously? 

Superhub Recommendation 

Future enterprise advantage will come from: 

• Process Readiness
• Data Readiness
• Governance Readiness 

not from deploying the highest number of AI tools. 

1. What is the biggest change in the Microsoft Teams 2026 update?

Teams is evolving from a collaboration tool into an AI-powered work platform, with AI becoming embedded across meetings, knowledge management, communication and daily workflows. 

2. What is an AI agent?

An AI agent is an intelligent system that can support multi-step tasks, retain context and help move work forward based on a defined business goal. 

3. What is the difference between Microsoft 365 Copilot and AI agents?

Microsoft 365 Copilot mainly helps employees improve personal productivity, while AI agents are more workflow-oriented and can support activities such as task follow-up, enquiry handling and process execution. 

4. Why is data governance important for AI adoption?

AI output depends on the quality, structure and permissions of enterprise data, so strong data governance is essential for reliable, secure and scalable AI adoption. 

5. Should Hong Kong SMEs pay attention to Agentic AI?

Yes. Hong Kong SMEs should start preparing identity management, access control and knowledge structures early, so future AI adoption is not limited by data or governance issues. 

The future of Microsoft Teams is no longer just about meetings and communication. 

More importantly, Teams is becoming a work platform that connects people, knowledge, workflows and AI. 

As AI evolves from a tool into a digital workforce that can participate in business workflows, organizations need to rethink not only technology deployment, but their overall operating model. 

For Hong Kong businesses, the next stage of AI advantage will not come from having the most AI tools. It will come from building an environment where people, knowledge, workflows and AI can collaborate securely and effectively. If your organization is evaluating Microsoft Teams, Microsoft 365 Copilot or AI agents, Superhub can help you design a practical AI adoption roadmap across process, data and governance readiness. 

Contact Superhub to explore how your organization can prepare for secure, scalable and well-governed AI-powered collaboration.

Source: What’s New in Microsoft Teams | June 2026 – InfoComm Edition | Microsoft Community Hub