What is the Learning agent in Microsoft 365 Copilot, and can it replace Copilot training?

Short answer: The Learning agent is a Microsoft agent inside Copilot that gives employees AI tips, skill assessments, role play practice and learning plans while they work, and Microsoft's 27 October 2026 Copilot release notes list it for Microsoft 365 Copilot on Windows. It is a useful way to keep Copilot skills growing after a rollout, but it doesn't replace practical training built around your own teams, data and rules.

What Microsoft has released

Microsoft's Microsoft 365 Copilot release notes for 27 October 2026, covering updates released between 13 and 27 October, include "Learning Agent in Microsoft 365 Copilot". Microsoft says it "helps employees build skills through personalized microlearning and structured learning experiences" and that it can help people "upskill on Copilot".

Microsoft's Learning agent overview lists what it can do:

• daily AI tips based on a person's role, to help them get more from Copilot
• task-based learning, with guidance and next steps tied to work
• AI Skills Navigator (in preview), which brings Microsoft's AI learning paths, sessions and credentials into one place
• AI-powered role play for practising workplace conversations
• assessments to check knowledge and find skill gaps
• short microlearning drawn from longer content, and structured learning plans
• connections to an organisation's own learning content, plus usage and engagement reports, where available

The same release notes add new Copilot metrics to the Copilot Dashboard and Advanced Analysis in Viva Insights. These cover actions in the Microsoft 365 Copilot app, Edge and OneNote, and features in Outlook, Word, Excel and PowerPoint such as suggested reply, translate, coach and clean data. Microsoft says this helps leaders "make informed decisions about training and deployment".

What you get depends on your licences

Microsoft's support page for the Learning agent says the experiences available "depend on the Microsoft products and licenses available in your organization". It splits them into three levels:

• Copilot Chat: AI tips, task-based learning, AI Skills Navigator, skill assessments and learning streaks
• Microsoft Copilot: adds personalised recommendations, prioritised microlearning and experiences grounded in SharePoint and OneDrive content
• Viva Learning: adds Copilot Academy, learning paths and content from learning management systems and supported third-party providers

Microsoft's setup guide says a Global Administrator deploys the agent from the Agents page in the Microsoft 365 admin center, to all users or chosen groups, and a Knowledge Administrator manages content sources. The guide lists Microsoft Copilot and Teams as supported, with Copilot Chat and side-by-side experiences in Word, PowerPoint and Excel marked as coming soon. Microsoft's older deployment FAQ, written for the preview, says LinkedIn Learning is only needed for role-play coaching scenarios.

Can it replace Copilot training?

It helps, but it is not a training programme on its own. The agent is strong at the small, frequent moments: a tip at the right time, a quick check of what someone knows, a pointer to a short video. What it can't do is decide which processes your teams should change, explain your AI policy, or show a finance team how to use Copilot on their real month-end reports.

UK research points the same way. The Skills for AI research on what works for AI upskilling was published on GOV.UK in June 2026 by DWP and Skills England. GOV.UK notes that the opinions in it are the author's and do not necessarily reflect the views of Skills England. Its employer guide says nearly all organisations report providing AI training (97%), yet many still identify gaps in flexibility (51%) and in practical, contextualised learning (34%). The research's PRIMES framework says good AI training should be practical, reachable, integrated, modular, expandable and sustainable.

The Learning agent fits the "reachable" and "modular" parts well, because it is in the flow of work and serves learning in small pieces. The "practical" and "integrated" parts still need someone to connect AI to your actual tasks, systems and governance. We think the strongest approach is to run hands-on, role-based sessions first, then use the agent to keep people practising afterwards.

Practical takeaways by role

L&D leaders: Treat the Learning agent as follow-up support, not the course itself. Plan what people will learn in live sessions, then decide which tips, assessments and learning plans should follow. Microsoft says you can connect SharePoint as a content source, so put your own prompt guides and process notes somewhere the agent can surface them. Our AI upskilling page explains how we build programmes around real work.

HR leaders: Personalisation draws on role information and skills signals, so check that job titles and skills data are accurate and that staff know what the agent uses. Microsoft's support page also says some experiences draw on SharePoint and OneDrive content, so explain clearly to staff what the agent uses before you roll it out. Our page on AI training for HR teams covers the people side of change, and our employee AI policy guide can help you set clear expectations.

IT leaders: Confirm which licences your people hold, because that decides which features appear. Agree who will deploy the agent, to which groups, and which content sources it can reach. Your Microsoft partner or internal IT team handles that setup. Our Microsoft Copilot training starts with a readiness check, so the training matches what people can actually see.

Transformation leaders: The new Viva Insights metrics give you a better view of which Copilot features are used and where adoption stalls. Take a baseline before training and look again a few weeks later. Our guide to measuring AI training ROI explains how to set that up, and our guide to business AI adoption covers the wider rollout.

A simple plan for the next month

1. Check access. Find out which licences your teams have and whether IT plans to deploy the Learning agent.
2. Take a baseline. Use the Copilot Dashboard to see current usage before you change anything.
3. Train the basics in person or live online. Use real tasks for each team, not generic demos.
4. Switch on follow-up. Point people to the agent's tips, assessments and learning plans for practice between sessions.
5. Review the data. Compare usage and confidence after a few weeks, and adjust the next round of training.

For more on building Copilot habits, our earlier post on skills in Copilot Chat is a good companion read, and the Copilot Essentials course suits people learning at their own pace.

What this means for your team

The Learning agent makes it easier for people to keep learning Copilot in small steps, and the new metrics make it easier to see whether that learning is turning into use. Neither tells your teams how AI should change their own work. Organisations that pair practical, role-based training with in-app follow-up and clear measurement are the ones most likely to see Copilot used well.

Not sure where your team stands today? Our free AI readiness scorecard takes a few minutes. To see how we help teams get value from Copilot, take a look at our Microsoft Copilot training. We also run sessions in person, including AI training in Leeds.

Talk to us

If you would like help planning Copilot training that works alongside the Learning agent, book a call with Day Seven.