Executive summary
- Microsoft is adding role-based Copilot agents for sales, service, and finance that sit inside the apps your teams already use, surfacing your business data to draft messages, summaries, and more.
- The agents will be available from October 2025 via the Copilot Agent Store for organisations that already use Microsoft 365 Copilot, so businesses can start small and scale as they start to rely on them more.
- The practical bit for UK teams is simple: run pilots on high-value tasks, put governance and data controls in place, and treat the rollout as a mini transformation project rather than a one-off installation.
Introduction
“Everything’s agentic”.
That’s the vision for Microsoft, at least – and they’re taking another step towards it with some new Copilot updates coming very soon.
From October 2025, Microsoft 365 Copilot will offer dedicated AI agents specifically built for certain business functions. More specifically: sales, service, and finance teams. In the announcement, Microsoft referenced “Frontier Firms” – organisations that are built with AI at the centre of how they work – and these agents are another building block for these companies.
The role-based agents will live in Outlook, Teams, and other Office apps, with the ability to pull data from CRM, ERP tools like Dynamics 365 and SAP, and help desk systems and use it to draft outputs like messages, summaries, and briefs.
These role-based agents will be available to install in the Microsoft 365 Copilot Agent Store at some point in October 2025.
What are the sales, service, and finance agents?
We all know that AI often comes with its fair share of hype, so let’s break this down to the bare essentials – just what are these agents and what do they do?
In short, they’re AI assistants which can do more than just chat – they’re specialised in working with other platforms (CRM, ERP, help desk, etc.) to pull data and surface it in the most useful manner.
Here’s how each one works:
- The sales agent handles meeting prep, pipeline prioritisation, and CRM housekeeping without all of the busywork. It can pull opportunities, review recent prospect communications, and bring multiple company signals together into a single brief. It can also update CRM records, generate next steps, and list at-risk deals for your sales team. In short, it helps them become more productive and efficient.
- The service agent operates similarly but its focus is on bringing down ticket churn. It can produce case summaries, draft customer-ready responses (which an agent can edit), and surface relevant knowledgebase articles automatically. The agent can also help prioritise work by suggesting any next actions, so your support team resolves issues faster.
- The finance agent is designed to tame spreadsheets and cashflow headaches. It brings ERP data into Copilot, Excel, and Outlook, and it can draft payment emails with account details, match transactions, and flag discrepancies. It’ll also manage variance analysis that highlights anything unusual, draft explanations, and clean ERP exports in tables that can be analysed.
What these agents all have in common is that they’re made to help teams get things done faster and more efficiently. And isn’t that the AI dream?
Is there a data protection concern here?
Since this brave new world of AI is still so new to so many of us, the thought of opening the door of your customer data to an AI might seem fraught with risk.
Thankfully, these agents are built on Microsoft’s world-class data protection controls – but, even beyond that, there are steps you should consider if you’re going to deploy agents in your business. You need to know what they can access, who can use them, and how you’ll track that activity.
In a real way, this is like a mini digital transformation for any business – paving the way for a more AI-centric future. So here are the steps we’d recommend:
- Decide which teams will pilot AI agents first and set clear success metrics, such as time saved on certain tasks or the total reduction in ticket handling times.
- Lock down data access and enable any available DLP (Data Loss Prevention) and audit logging to ensure you can see what the agent reads and outputs.
- Create review and approval workflows for customer-facing or finance-critical drafts, so responsibility always stays with your people. Human-first is a solid approach to any AI transformation.
- Finally, train users on prompt hygiene and how to recognise when to trust an AI draft vs. when human oversight is essential.
Need a hand with your AI transformation?
This latest slew of AI agents come with a practical promise: do the repetitive stuff faster, and let your teams spend their time and energy on more important things.
For UK businesses, now is the time to pilot, measure, and scale AI-powered solutions across their organisations. If you need any assistance with that, just reach out to your Get Support Customer Success Manager or call our team today on 01865 594 000.