Microsoft Copilot Review 2026

Microsoft Copilot is one of the strongest AI purchases in 2026 for organisations that already depend on Microsoft 365. Its advantage is not that it is a universally better chatbot than ChatGPT, Claude or Gemini. It is that AI can sit inside Word, Excel, PowerPoint, Outlook and Teams and work within the same identity, permissions and productivity environment employees already use.

That ecosystem fit makes the buying decision unusually context-dependent. A company already paying for Microsoft 365 can gain meaningful value from reducing copying between a separate AI assistant and its documents, spreadsheets, email and meetings. A Google-first or mixed-stack company may get less return and could be better served by Gemini, ChatGPT or Claude.

For organisations, current UK pricing lists Microsoft 365 Copilot at £23.10 per user per month when paid yearly, excluding VAT, with a separate qualifying Microsoft 365 licence required. Microsoft also offers small-business bundles and a Copilot Business offer whose pricing varies by billing cycle and whether Teams is included. The true cost is therefore the combined licence stack, not a single headline figure.

Microsoft Copilot is better understood as an AI layer across Microsoft's software than as one isolated chatbot. The Copilot name covers consumer experiences, Copilot Chat, Microsoft 365 Copilot, business bundles, agents and Copilot Studio, as well as AI embedded across Windows and other Microsoft products.

That breadth is powerful but confusing. Two users can both say they 'have Copilot' while using materially different products with different access to company data, Office applications, administration and usage limits. Any review that gives one price or one feature list without specifying the plan is incomplete.

For this review, the central product is Microsoft 365 Copilot for work: the version designed to connect generative AI with Microsoft 365 applications and organisational context. We also cover Copilot Chat and small-business licensing because those determine how companies actually buy the product.

ToolRankers did not run a proprietary benchmark suite. The review is based on current Microsoft pricing, product, privacy and security documentation plus current independent market coverage reviewed on 7 October 2026.

Copilot is most compelling for organisations where Microsoft 365 is already the operating environment. If employees spend their day in Outlook, Teams, Excel, Word and PowerPoint, the assistant can reduce context switching because it appears closer to the work.

It is particularly relevant to managers, analysts, sales staff, consultants, finance teams, administrators and knowledge workers who spend meaningful time reading, drafting, summarising and transforming information inside Microsoft applications.

Developers and creative professionals can still benefit, but they may have more specialised alternatives. Claude or ChatGPT may offer a more focused coding or general AI workspace, while dedicated creative tools can be better for advanced design production.

A company not already committed to Microsoft 365 should not choose the entire ecosystem solely to gain Copilot. AI is one factor in productivity-suite selection, not a sufficient reason to migrate email, identity, files and collaboration on its own.

Standalone assistants are powerful but impose a transfer cost. A user copies an email into the chatbot, asks for a draft, moves the answer back into Outlook and then repeats the process with a spreadsheet or document. Copilot's strategic promise is to remove part of that transfer because the assistant is embedded within the productivity suite.

This matters more than it sounds. Small frictions multiplied across hundreds of employees and thousands of tasks can outweigh a modest difference in model quality. If an assistant has the right document, meeting or spreadsheet context without manual preparation, the time saved begins before generation starts.

The integration advantage also affects training. Employees already know what Word, Excel and Outlook are for. Instead of asking them to adopt a blank general-purpose AI workspace, the organisation can teach concrete tasks inside familiar applications.

The trade-off is ecosystem dependence. The closer Copilot becomes to Microsoft 365, the less value it provides to a company whose work lives elsewhere.

Microsoft's licensing structure is one of the hardest parts of evaluating Copilot. In the UK, the enterprise Microsoft 365 Copilot add-on is currently listed at £23.10 per user per month when paid yearly, excluding VAT. A qualifying Microsoft 365 licence is required, so that is not the full per-user software cost.

Microsoft also sells business bundles aimed at smaller organisations. Current UK pages show Business Standard with Copilot and Business Premium with Copilot at different annual prices depending on whether Teams is included. A separate Microsoft 365 Copilot Business offer has also been promoted at a lower add-on price than the enterprise product. These prices and promotions can change, which makes live verification essential at purchase time.

The correct cost comparison therefore includes the base Microsoft 365 plan, the Copilot entitlement, Teams configuration, annual versus monthly billing and any agent or metered usage that sits outside the core licence.

A business should build a role-based cost model rather than multiply one advertised number by its headcount. Employees who rarely use the productivity applications may not justify a premium Copilot licence, while a finance analyst or account manager could save enough time to make the same seat inexpensive.

Microsoft offers Copilot Chat to eligible users with Microsoft Entra accounts at no additional cost in many qualifying Microsoft 365 environments. This gives organisations a lower-friction way to expose employees to enterprise-oriented AI chat before purchasing the deeper Microsoft 365 Copilot experience for everyone.

The important distinction is work grounding and in-app integration. Paid Microsoft 365 Copilot is designed to use Microsoft 365 context and appear more deeply inside the productivity applications, subject to the user's permissions and the capabilities available in the plan.

This creates a useful deployment path. A company can make Copilot Chat available broadly for lower-risk general use and assign paid Copilot licences to roles where access to work data and in-app features has measurable value.

The mistake is assuming the free chat experience proves the ROI of the paid product. The most valuable paid workflows depend on the very integration that Copilot Chat does not reproduce completely.

For small businesses, Microsoft offers combinations of Microsoft 365 Business plans and Copilot rather than forcing every company into the enterprise licensing path. The commercial appeal is bundling: productivity software, security features and AI can be purchased within one vendor relationship.

Business Standard with Copilot is suited to organisations that need desktop Office apps and collaboration, while Business Premium with Copilot adds stronger security and device-management capabilities around the productivity environment. The exact bundle economics depend on Teams inclusion, billing term and promotional pricing.

Small companies should not automatically buy Copilot for every employee. Start with roles where Microsoft 365 activity is high: sales, finance, operations, management, client services and administration. Measure specific tasks before expanding seats.

The best SMB case is not 'everyone gets AI'. It is 'the people who spend hours in Microsoft 365 get an assistant inside the software that already contains their work'.

Enterprise deployment adds a different set of considerations: identity, compliance, information architecture, permissions, data residency, eDiscovery, auditing and the cost of hundreds or thousands of seats. At that scale, a small per-user productivity improvement can be valuable, but an unused licence becomes expensive very quickly.

Microsoft positions Copilot inside its existing enterprise security and privacy framework. The assistant works with the user's permitted Microsoft 365 context rather than creating a separate uncontrolled data silo. This is a major advantage for companies that already govern documents, email and collaboration through Microsoft.

However, Copilot can make existing permission problems more visible. An employee may technically have access to a forgotten SharePoint folder and, with AI search, discover information they would never have found manually. The correct response is not to blame Copilot; it is to fix access governance before broad deployment.

Enterprise buyers should therefore treat content and permission readiness as part of the implementation project, not as a separate housekeeping task.

Word is one of the clearest use cases because documents are both time-consuming and common. Copilot can help draft, rewrite, summarise and transform material without forcing the user to leave the document environment.

The business value appears in first drafts and revision loops. A consultant can turn notes into a proposal structure, a manager can summarise a lengthy report, and an administrator can rewrite a policy section for clarity. The time saved is often in getting from a blank page to a workable draft.

The limitation is accountability. A fluent draft can still contain factual errors, unsupported claims or inappropriate tone. Legal, policy and client-facing documents need review by someone who understands the subject.

Microsoft's recent emphasis on source citations inside Word also reflects the industry's broader recognition that professional AI needs traceability, not only polished prose. Citations are useful, but the user still needs to inspect the underlying material.

Excel may be the highest-value Copilot surface for non-specialist users because spreadsheets combine business knowledge with technical syntax. A manager can know what question needs answering without knowing the correct formula, pivot table or data transformation.

Copilot can help users explore data, explain formulas, generate calculations and identify patterns. That can make analytical work more accessible and reduce the number of basic spreadsheet questions escalated to specialists.

The risk is that AI can make incorrect analysis look authoritative. Poorly labelled columns, inconsistent dates, missing values or a misunderstood business definition can produce a plausible but wrong conclusion. The more financially significant the decision, the stronger the verification requirement.

For finance and operations teams, the best practice is to use Copilot to accelerate exploration and documentation while preserving normal controls around reconciliations, approvals and material financial outputs.

PowerPoint integration is useful because presentation creation is often a transformation task: a report, meeting or set of notes has to become a shorter narrative with slides. Copilot can accelerate that first pass.

The time saving is most obvious for people who create decks frequently but are not presentation specialists. Sales teams, managers and consultants can use AI to create an initial structure, rewrite slide text and generate a draft that is easier to refine than a blank presentation.

The final deck still needs human judgement. Good presentations depend on emphasis, sequencing, visual hierarchy and knowing what to remove. AI can create too much content as easily as too little.

For high-stakes pitches, board meetings or external presentations, Copilot should be treated as a production accelerator, not an autonomous presentation designer.

Email is a natural AI target because it consumes time in small increments throughout the day. Copilot can help summarise threads, draft replies and reduce the effort required to process long conversations.

The value is particularly high for customer-facing roles and managers with overloaded inboxes. A faster understanding of a thread can reduce response time and make it easier to identify decisions or outstanding actions.

Email is also a high-risk channel for automation because an incorrect message can leave the organisation immediately. Drafting should not be confused with sending. Sensitive customer, HR, legal or financial communication needs a human review step.

Businesses should train employees to verify names, numbers, commitments and tone before sending AI-assisted email, especially when the assistant has summarised a long thread that the user did not read in full.

Teams is where Copilot can turn meeting content into a more searchable work asset. Depending on configuration and licensing, users can ask questions about discussions, catch up on what they missed and extract actions or themes from meeting context.

This can reduce the cost of internal coordination. Employees who miss a meeting may not need another meeting simply to learn what happened, and managers can spend less time manually producing summaries.

Meeting AI also creates governance responsibilities. Organisations need clear policies around recording, transcription, retention and who is allowed to access the output. Sensitive meetings should not be treated like routine status calls simply because automatic summarisation is convenient.

The feature is most valuable in companies that already use Teams consistently. A business that meets elsewhere will not capture the same return.

Microsoft increasingly describes Copilot's value through its ability to work with organisational context and what it calls Work IQ. The idea is that the assistant becomes more useful when it can understand the user's work, relationships, files and processes within the permissions of Microsoft 365.

This is strategically important because generic model intelligence is becoming easier to access. What distinguishes an enterprise AI system is often whether it can ground answers in the organisation's real information without forcing users to manually assemble context every time.

The benefit grows with information quality. If SharePoint sites are current, files are named well and permissions reflect real responsibilities, work grounding can save significant time. If the environment is chaotic, Copilot can surface chaos efficiently.

Information architecture therefore becomes part of AI readiness. Companies should see Copilot adoption as a reason to improve data hygiene, not as a substitute for it.

Search is an important but less glamorous part of the Copilot value proposition. Knowledge workers waste time looking for documents, messages and decisions that already exist. An AI layer can make retrieval more conversational and reduce the dependency on exact filenames or folder locations.

Notebooks and related organisational features can help users collect context around a project rather than repeatedly reconstructing it. The value is continuity: a project assistant becomes more useful when it can work from an organised body of material rather than isolated prompts.

The same caution applies as elsewhere. Search results reflect permissions and the quality of the underlying content. AI does not automatically know which of three conflicting policy documents is the current one unless the organisation maintains that distinction.

For companies with large Microsoft 365 estates, search and context may ultimately create more day-to-day value than headline generative features.

Microsoft is pushing Copilot beyond chat toward agents that can perform structured work. Copilot Studio allows organisations to build and manage agents connected to business processes and data sources, and some agent usage can introduce metered costs beyond the standard licence.

This is where the potential productivity gains become larger—and the risks become more operational. An assistant that drafts a summary can be checked easily. An agent that updates records, sends messages, changes a workflow or acts across systems can create consequences before anyone notices an error.

Businesses should begin with narrow agents whose purpose, permissions and success criteria are clear. High-impact actions should require approval, and logging should make it possible to understand what the agent did.

The right question is not 'How autonomous can we make Copilot?' but 'Which repetitive decisions can safely be delegated while keeping accountability with a human owner?'

Microsoft Copilot uses leading generative models, including OpenAI technology, but buyers should not treat it as a fixed wrapper around one model version. Microsoft can change the model mix, routing and product features as the platform evolves.

That matters because some users compare Copilot with ChatGPT as though the underlying model automatically makes them equivalent. The surrounding product is different. Microsoft adds work grounding, identity, application integration and enterprise controls; OpenAI's own ChatGPT adds a different set of tools, agents, apps and workspace features.

For organisations, model quality is one component of the purchase rather than the whole product. A slightly stronger model that cannot access the right company context may produce less value than a well-integrated assistant with appropriate permissions.

Do not buy Copilot solely because of today's model name. Buy it if the Microsoft platform around the model solves the workflow.

Microsoft's work and consumer AI experiences have different data-handling contexts. Microsoft states that Microsoft 365 Copilot operates within the Office 365 trust boundary and that organisational files and communications are not used to train foundation models. Microsoft also states that Microsoft Graph data is not used to train foundation models.

Consumer Copilot has separate privacy controls. Microsoft has disclosed circumstances in which signed-in consumer conversation activity may be used for model improvement unless the user opts out, with exclusions for categories such as organisational Entra users and Copilot inside Microsoft 365 consumer apps.

This is why businesses should not build a data policy around the word 'Copilot' alone. The account type, product surface and licence matter. An employee pasting confidential client information into a personal consumer experience is a different risk from using an approved organisational Copilot environment.

Procurement and IT teams should document which Copilot surfaces are approved for which categories of data and train employees to recognise the difference.

Microsoft's strongest enterprise argument is that Copilot inherits the security, privacy and compliance controls of the Microsoft environment rather than bypassing them. This can simplify deployment for organisations already invested in Microsoft identity and compliance tooling.

But inherited permissions are only as good as the permissions themselves. Years of broad SharePoint links, legacy Teams membership and poorly maintained groups can create oversharing. Copilot can make content easier to discover because users no longer need to know exactly where it lives.

Before rollout, organisations should audit highly sensitive repositories, review broad access groups, establish data classification and remove obsolete content where appropriate. Copilot readiness is partly information-governance readiness.

The goal is not to restrict useful knowledge unnecessarily. It is to ensure that AI makes authorised information easier to use without making accidental access easier to exploit.

Copilot's greatest usability advantage is familiarity. Users already understand the purpose of Word, Excel and Outlook, so the AI can be introduced through tasks they recognise rather than abstract prompting lessons.

The challenge is discoverability and inconsistency across plans. Features can appear in different places depending on licence, account type, application and rollout. Employees may own Copilot without understanding which workflows justify opening it.

Training should therefore be role-based. A salesperson needs examples around account research, follow-up and proposals; a finance analyst needs spreadsheet exploration and commentary; a manager needs meeting and document workflows. Generic 'how to prompt AI' sessions are less effective than examples tied to real work.

Adoption should also be measured. Licence assignment is not usage, and usage is not value. The organisation needs evidence that people are completing meaningful tasks more efficiently.

Sales

Measure time spent preparing account briefs, drafting follow-ups, summarising meetings and creating proposals. Copilot has strong potential when CRM and Microsoft 365 information can be brought together safely.

Finance

Measure time spent explaining formulas, preparing commentary, checking spreadsheet trends and turning data into management material. Keep normal financial controls around important outputs.

Management

Measure meeting catch-up, document review, presentation preparation and internal communication. The risk is accepting summaries without reviewing the evidence behind important decisions.

Operations and administration

Measure repetitive document creation, email processing, meeting actions and information retrieval. These roles can benefit substantially because work is often high-volume and process-oriented.

Technical teams

Compare Copilot with specialist coding assistants rather than assuming the Microsoft 365 licence solves developer needs. The best tool for code may be different from the best tool for Office productivity.

A useful pilot begins with a limited group of employees whose work already creates a plausible ROI. Choose three to five workflows, capture the existing time and quality baseline, and document what data the participants are allowed to use.

During the pilot, collect failure examples as aggressively as success examples. Which summaries omitted important context? Which spreadsheet analyses required correction? Which employees stopped using the tool after the first week? Adoption problems often reveal more than enthusiastic testimonials.

Review Microsoft 365 permissions before enabling the deepest work-grounded use. A pilot is also a good time to identify obsolete files, broad groups and repositories that need governance work.

At the end, compare seat cost with observed savings. Expand licences to roles with clear value and leave lower-use employees on Copilot Chat or existing tools until a stronger use case emerges.

Five-person professional-services firm

If everyone already uses Business Standard and spends substantial time in Word, Outlook and Teams, a Copilot bundle may be simpler than purchasing a separate general assistant for each person. The company should still test whether all five roles need the paid integration.

Twenty-person company with only five heavy knowledge workers

Licence the five high-value roles first. Broad Copilot Chat availability can cover lighter general needs while avoiding 15 additional premium seats with unclear ROI.

Security-conscious small business

Business Premium with Copilot can be attractive because the decision is not only about AI; stronger identity, device and security capabilities may be part of the bundle value. Compare the full Microsoft 365 requirement, not the Copilot component in isolation.

Google-first company

Do not buy Microsoft 365 Copilot merely because it is capable. If employees work in Gmail, Docs and Sheets, Gemini will usually have the stronger integration case unless the company already has another strategic reason to move to Microsoft.

ChatGPT is the stronger general-purpose choice for users who want an independent AI workspace spanning research, files, coding, creative tasks, voice and connected apps. It does not assume that Microsoft 365 is the centre of the workday.

Copilot is stronger when the work already lives in Microsoft applications. An employee drafting in Word, analysing in Excel and communicating through Outlook or Teams can benefit from native context and fewer transfers.

For many businesses the tools are not exact substitutes. Copilot can be the organisational productivity layer while specialist teams maintain ChatGPT access for broader or agentic work. The second subscription should have a defined role rather than existing because both products are popular.

For more detail, read our ChatGPT review.

Copilot and Gemini are increasingly parallel ecosystem strategies. Microsoft embeds AI across Microsoft 365 and Windows; Google embeds Gemini across Gmail, Docs, Sheets, Drive, Search and Android.

The existing productivity suite should usually decide the shortlist. Migrating thousands of files, email accounts and workflows merely to access a preferred AI assistant is rarely rational when both ecosystems have capable AI.

Gemini may appeal more to Google-first organisations and users who value very large context windows; Copilot is the more natural choice for Word, Excel, Outlook and Teams-centred work.

For more detail, read our Google Gemini review.

Claude is a more focused AI workbench for writing, coding, documents and sustained analytical tasks. Copilot is a more integrated productivity layer for Microsoft applications.

A developer or writer may prefer to spend long sessions inside Claude. A finance analyst or operations manager may get more value from Copilot because the spreadsheet, email and meeting context already sits in Microsoft 365.

The better product depends on where the expensive part of the work occurs. If it happens inside the assistant, Claude can win; if it happens inside Office applications, Copilot has the stronger fit.

For more detail, read our Claude review.

• Deep integration with Word, Excel, PowerPoint, Outlook and Teams.
• Strong fit for organisations already standardised on Microsoft 365.
• Enterprise-oriented identity, privacy, security and compliance positioning.
• Reduces context switching by bringing AI into existing applications.
• Copilot Chat provides a lower-cost entry point for eligible users.
• Agents and Copilot Studio extend the product beyond drafting into workflows.
• Can use organisational context subject to existing permissions.

Copilot's moat is not one model. It is the combination of applications, data, identity and enterprise controls that Microsoft already owns.

• Licensing is complex and the headline Copilot price may exclude the qualifying Microsoft 365 plan.
• Value falls sharply for businesses that do not already use Microsoft 365 heavily.
• Existing permission problems can become easier to discover through AI.
• Feature availability can vary by plan, application and rollout stage.
• Agent usage and other advanced capabilities can create additional costs.
• Some specialist writing or coding users may prefer Claude or ChatGPT.
• Broad seat deployment can become expensive if adoption is weak.

The key compromise is dependence on Microsoft. The integration that makes Copilot valuable also makes the product less compelling outside the ecosystem.

A business that primarily uses Google Workspace should usually evaluate Gemini first. A developer-heavy team that spends little time in Microsoft 365 may get more value from Claude, ChatGPT or specialist coding tools.

Companies with poor information governance should also avoid rushing into broad work-grounded deployment. Clean up sensitive permissions before making organisational knowledge dramatically easier to query.

Finally, do not licence every employee simply because the company negotiated a good per-seat price. Unused software is not a productivity investment. Match licences to workflows and expand based on evidence.

Copilot can be excellent value for roles that live inside Microsoft 365. A recurring monthly seat cost is easy to justify when the employee consistently saves hours across documents, spreadsheets, email and meetings.

The value is weaker when usage is occasional or when most work happens outside Microsoft applications. That is why role-based deployment is more sensible than an automatic company-wide purchase for many small and mid-sized organisations.

For procurement, calculate total licence cost and include the qualifying Microsoft 365 subscription, any Teams configuration and expected agent usage. Compare that total with observable time savings and software it may replace.

The strongest reason to buy Microsoft Copilot is integration with an existing Microsoft environment. The most important compromise is that you pay for that ecosystem fit even if some standalone AI competitors are stronger on particular tasks.

The enterprise AI market is shifting from assistant demos toward measurable operational value. Microsoft has enormous distribution, but buyers increasingly want evidence that Copilot produces sustained productivity gains after the novelty wears off.

The company is also broadening Copilot toward an 'OS for work' model with richer agents, coding and integrated productivity experiences. That direction can make the product more valuable, but it also increases complexity and the importance of security controls around autonomous actions.

For customers, the correct response is neither scepticism nor hype. Treat new capabilities as potential workflow improvements, test them on real tasks and expand only when the value survives normal day-to-day use.

Copilot's long-term success will depend less on whether it can generate impressive text and more on whether Microsoft can make AI reliably useful inside the systems companies already trust to run their work.

Microsoft Copilot is the best AI purchase for many organisations already committed to Microsoft 365 because its integration advantage is real. Word, Excel, Outlook, PowerPoint and Teams are not peripheral tools in those companies; they are where the work happens. Bringing AI closer to that context can remove more friction than choosing a separate assistant on model quality alone.

The product is not an automatic company-wide recommendation. Licensing is complex, the total cost includes the surrounding Microsoft 365 environment, and ROI varies sharply by role. Pilot high-use employees first, clean up permissions, measure real workflows and expand only where the seat saves observable time or improves output.

The strongest reason to choose Copilot is Microsoft 365 integration. The most important compromise is ecosystem dependence. If your company is not already Microsoft-centric, ChatGPT, Claude or Gemini can offer more flexibility and may provide better value.

Is Microsoft Copilot free?

Microsoft offers Copilot Chat at no additional cost for many eligible Microsoft Entra users with qualifying Microsoft 365 access. Deeper Microsoft 365 work grounding and in-app features require paid Copilot licences or bundles.

How much is Microsoft 365 Copilot in the UK?

The current enterprise listing is £23.10 per user per month when paid yearly, excluding VAT, with a separate qualifying Microsoft 365 licence required. Small-business bundle pricing differs.

Do I need Microsoft 365 for Copilot?

Consumer Copilot experiences can be used separately, but the deepest Word, Excel, Outlook, Teams and organisational integrations depend on Microsoft 365 plans and Copilot licensing.

Is Copilot better than ChatGPT?

Copilot is usually better integrated for Microsoft 365 work. ChatGPT is a broader independent AI workspace. The better choice depends on whether Microsoft 365 is the centre of your workflow.

Can Copilot analyse Excel spreadsheets?

Yes, on eligible plans Copilot can assist with Excel analysis, formulas and understanding spreadsheet data. Important financial conclusions should still be independently checked.

Does Microsoft use my company data to train AI models?

Microsoft states that Microsoft 365 Copilot operates within the Office 365 trust boundary and that organisational files, communications and Microsoft Graph data are not used to train foundation models.

Can Copilot access everything in my company?

No. Copilot works within the user’s existing permissions. However, overly broad legacy permissions can make information easier to discover, so organisations should review access before rollout.

Is Microsoft Copilot worth it for a small business?

It can be excellent value for a Microsoft 365-centred business with employees who spend substantial time in Word, Excel, Outlook and Teams. It is less compelling for low-use or non-Microsoft roles.

Should every employee get a Copilot licence?

Not initially. A role-based pilot is usually better. Assign paid seats where work-grounded AI and in-app integration have a measurable use case, and expand after observing value.

What is the biggest risk with Copilot?

For many organisations, the biggest practical risk is not the model itself but poor data governance: existing oversharing and excessive permissions can become easier to surface through AI.

ToolRankers checked these primary and independent sources during research. Re-check time-sensitive pricing, limits and policies before a later republication date.

• Microsoft — Microsoft 365 Copilot pricing — https://www.microsoft.com/en-gb/microsoft-365-copilot/pricing
• Microsoft — Microsoft 365 Copilot enterprise pricing — https://www.microsoft.com/en-gb/microsoft-365-copilot/pricing/enterprise
• Microsoft — Microsoft 365 Copilot business pricing without Teams — https://www.microsoft.com/en-gb/microsoft-365-copilot/pricing/no-teams
• Microsoft — Copilot in apps for work — https://www.microsoft.com/en-gb/microsoft-365-copilot/in-apps-for-work
• Microsoft Learn — Data, privacy and security for Microsoft 365 Copilot — https://learn.microsoft.com/en-us/copilot/microsoft-365/microsoft-365-copilot-privacy
• Microsoft Learn — Microsoft Graph and Copilot data protection — https://learn.microsoft.com/en-us/copilot/microsoft-365/microsoft-365-copilot-privacy
• Microsoft Support — Copilot privacy FAQ — https://support.microsoft.com/en-us/topic/copilot-privacy-faq-4036e11f-5fa6-4a72-9d8a-6750b2e17375
• The Verge — Inside Microsoft’s Copilot rethink — https://www.theverge.com/tech/1003365/microsoft-copilot-os-for-work-notepad