Microsoft Copilot Studio Agent Development UK.

What Copilot Studio is good for

Microsoft Copilot Studio is Microsoft's tool for building AI agents that run in Teams, Microsoft 365 Copilot Chat, SharePoint, websites and other channels. It sits inside the Power Platform, so it shares environments, connectors and governance with Power Apps and Power Automate.

It is a strong choice when your knowledge and users already live in Microsoft 365. Typical agents we would build:

• An internal helpdesk agent that answers HR, IT and policy questions from SharePoint, with links to the source.
• A sales or service agent that answers product questions and checks order status in your ERP through a connector.
• A process agent that collects details for a request, such as a holiday booking or a returns authorisation, and starts the approval flow.

If your organisation is not on Microsoft 365, or you need full control of the model and interface, a custom LLM application is usually a better fit, and we will say so.

Grounding agents in SharePoint and Dataverse

Microsoft Learn's knowledge sources summary lists public websites, documents uploaded to Dataverse, SharePoint, Dataverse tables and enterprise data through Copilot connectors. For SharePoint, Dataverse and connector sources, the agent authenticates as the person asking, so it only surfaces content that user can already access. That is the right model, but it also means messy SharePoint permissions become visible very quickly.

Most poor answers come from the content, not the agent: five versions of the same policy, scanned PDFs with no text layer, or libraries that mix current and archived documents. Part of our work is tidying the sources the agent uses and pointing it at specific sites and libraries rather than everything.

Copilot Studio has an Allow ungrounded responses setting. With it off, Microsoft says the agent blocks replies that did not use a knowledge source or tool. For policy and compliance agents we normally turn it off and write instructions that require citations. See RAG knowledge bases for the same ideas outside Microsoft's platform.

Connecting to systems outside Microsoft 365

Agents become useful when they can look things up and act. Copilot Studio offers three main routes, and we choose per system:

• Power Platform connectors: hundreds of prebuilt connectors, such as Dynamics 365, Salesforce and SQL Server, many of them premium.
• Custom connectors built from an OpenAPI definition of your own API, with OAuth 2.0 or API key authentication.
• MCP servers: Microsoft Learn documents an onboarding wizard for connecting an existing MCP server. It requires generative orchestration and supports the Streamable transport only; Microsoft says SSE has not been supported since August 2025.

Microsoft Learn also notes that MCP access runs through Power Platform connectors, so a data policy that blocks a connector blocks the MCP tools behind it. That is useful: your existing governance applies. For multi-step work we use agent flows in Power Automate, with human approval steps for anything that changes records or sends messages to customers.

Licensing and Copilot Credits, from Microsoft's own pages

Licensing is where most Copilot Studio projects stall, so we work it out early, using Microsoft's documents rather than memory. As of Microsoft Learn's licensing page updated in November 2026:

• Copilot Credits are the common currency. They replaced messages on 1 September 2025.
• Credits come from pay-as-you-go billed through an Azure subscription, prepaid Copilot Credit pack subscriptions, or a one-year prepurchase plan.
• Makers need a Copilot Studio user licence, which is free but requires the tenant to hold a prepaid credit pack, or a Microsoft 365 Copilot licence. A trial lets you build and test but not publish.
• Microsoft says users with a Microsoft 365 Copilot licence are not charged for classic answers, generative answers or Microsoft Graph grounding when they use agents in Copilot Chat, Teams or SharePoint under their own identity.
• Unused prepaid credits do not carry over month to month, and exceeding purchased capacity can lead to service denial.

Microsoft Learn now also describes more than one agent "harness", and licensing differs between them, so we confirm which applies to your agent. Microsoft publishes an agent usage estimator and a licensing guide; we use both to forecast credits before you commit. We do not quote Microsoft's prices on this site because they change; check Microsoft's pricing page or your licensing partner.

Testing, governance and handover

Before launch we build a set of real questions with known good answers and run it after every change to instructions, knowledge or tools. We set up separate development and production environments, put the agent in a Power Platform solution so it can be moved cleanly, and check data policies with your IT team. After go-live, analytics show which questions fail and where credits go. We hand over written admin notes so someone in your business owns the content. Every Copilot Studio project is a fixed-price quote after a free chat.

What we deliver

• A scoped agent design: audience, channel, knowledge sources, tools and what it must refuse
• Knowledge setup across SharePoint, Dataverse and documents, with content tidied where needed
• Custom connectors or an MCP server for systems outside Microsoft 365
• Agent flows in Power Automate for multi-step tasks, with approvals where needed
• A test set of real questions with expected answers, run before every change
• Copilot Credits estimates, usage monitoring and alerts
• Admin notes covering data policies, environments and who maintains what

How it works and what it costs

Every project gets a fixed-price quote after a free initial chat and a short scoping stage. You own the code and the data.

1. Free chat

   Tell us the problem in plain English: what you do now, what goes wrong and what "better" looks like. No charge, no obligation.

2. Scoping

   We map the processes, systems and data involved, agree what is in and out, and write it down so there are no surprises.

3. Fixed-price quote

   You get a fixed price for the agreed scope, or a phased plan for bigger builds, so you can start small and prove it works.

4. Build and test

   We build in short stages you can see and try, test against real data, then go live carefully with a rollback plan.

5. Hand over and look after

   You own the code and the data. We can host it, support it and keep improving it, or hand it to your own team.

Related