Microsoft Copilot has quietly become five products in a trench coat

Open Microsoft 365 Copilot today and you'll find something that looks nothing like the chatbot that launched a couple of years ago. There's a chat window, sure. But there's also Researcher, Analyst, Notebooks, Pages, an Agent Store, and now Cowork — a system that can run tasks in the background while you do something else entirely. Somewhere in that pile is probably the exact tool that would save you three hours this week, if you knew it existed.

Most people never find it. They open Copilot, ask it to summarize an email, and close the tab. Meanwhile, Copilot has quietly turned into something closer to an operating system for AI work — one where the chat box is just the front door.

This guide walks through what's actually in that ecosystem now, what each piece is for, and how to move from "asking Copilot a question" to "handing Copilot a job."

What is Microsoft 365 Copilot actually made of now?

Think of Copilot less as one tool and more as a layer that sits across your Microsoft apps and connects to several distinct experiences. You can access the whole thing at m365.cloud.microsoft or through the Microsoft 365 Copilot app.

The core pieces, according to Microsoft's own product breakdown, are: bringing together your Copilot chats, files, meeting notes, and project materials in Notebooks, plus adding agents to Copilot that automate common tasks or work on your behalf. On top of that, Microsoft now offers four ways to organize work with Copilot: Copilot Notebooks, Copilot Pages, agents and Copilot Cowork, and they solve different things, with the right choice depending on what you want to achieve.

That's the shift worth understanding before anything else: Copilot isn't one interface anymore. It's a set of tools, and picking the wrong one for the job means you're doing extra work for nothing.

How is Copilot Chat different from a regular chatbot?

The chat window itself got a major upgrade you might have missed: model choice. You're no longer locked into a single AI brand. You don't need to switch between AI model providers for the best models for work — Copilot has options from Anthropic and OpenAI available directly from Microsoft 365. GitHub's own Copilot documentation confirms this multi-model approach extends further, with developers able to choose between multiple large language models, including OpenAI's GPT models, Anthropic's Claude, xAI's Grok, and Google's Gemini.

Why does this matter practically? Because different models are genuinely better at different things — one might be sharper at writing, another at reasoning through a spreadsheet formula. Instead of paying for four separate subscriptions and copy-pasting between tabs, you pick the model inside the same window and keep your context intact.

There's also an "Auto" option if you don't want to think about it. In the GitHub Copilot ecosystem, Auto model selection automatically chooses the best available model for your prompts, reducing the mental load of picking a model and helping you avoid rate-limiting. Microsoft 365 Copilot follows the same logic — let it pick, or override it when you know exactly which model suits the task.

What's the difference between Researcher and just asking Copilot a question?

This is where things get genuinely interesting, and where most casual users never venture. Researcher is not a smarter version of chat — it's a fundamentally different tool built for a different kind of problem.

Microsoft's own documentation draws a sharp line here: the standard Copilot experience helps with a broad range of everyday tasks across Microsoft 365 apps, optimized for speed and efficiency, great for quicker tasks like summarizing emails or drafting short replies. Researcher works differently. Researcher agent is designed for deeper reasoning of your tasks, excelling at tackling complex tasks that require thoughtful analysis, taking longer processing time to deliver a more comprehensive, insightful response.

The way it investigates is deliberately modeled on how a person would tackle the same job. Microsoft's engineering team describes the approach this way: the methodology mirrors what a human would do when researching a subject — seek clarification, devise a higher-order plan, break the problem into subtasks, then begin an iterative loop of Reason → Retrieve → Review for each subtask, collecting findings until synthesizing the final report.

In practice, that means you can hand Researcher a messy client brief and it will pull out details you didn't explicitly repeat — timelines, target markets, budget constraints — because it's actually reading and reasoning over the document, not just pattern-matching keywords.

To try it yourself:

1. Sign into the Microsoft 365 Copilot app
2. Find Researcher under the Agents section in Chat — you can find Researcher under Agents in Chat after signing into the Microsoft 365 Copilot app
3. Attach a file or email thread by typing "/" before the prompt
4. Answer any clarifying questions it asks — Researcher often asks a couple of questions to focus the work; answer them, or just say "Go ahead"
5. Let it run, then use the output — you can edit the report, share it, or use it to kick off a presentation or document

Microsoft's official Researcher documentation is worth bookmarking, and the Researcher FAQ on Microsoft Learn answers most of the "wait, can it access X" questions you'll have.

What happened to Deep Research?

If you used Copilot's older "Deep Research" feature, it's gone now — but not lost. Researcher and Deep Research were very similar experiences that helped customers create detailed research reports and analyses; as part of ongoing updates to Copilot, Deep Research has been retired and Researcher is now the in-depth research experience. Your old reports didn't disappear either — existing Deep Research reports are not deleted, and Microsoft 365 Premium and Pro subscribers can continue accessing their previous reports in Researcher.

Should you use Analyst instead of Researcher?

If Researcher is your investigator, Analyst is your data person. Both launched together as what Microsoft called two first-of-their-kind reasoning agents for work that analyze vast amounts of information with secure, compliant access to your work data — your emails, meetings, files, chats, and more — and the web to deliver highly-skilled expertise on demand.

The distinction matters when you're deciding which one to open. Researcher writes reports and cites sources. Analyst is built to sit inside spreadsheets and work through numbers — think of it as the tool you reach for when the question is "what does this data actually show" rather than "what do I need to know about this topic."

What is Copilot Notebooks and how is it different from OneNote?

Notebooks solves a problem everyone with a messy desktop understands: project material scattered across ten different files and three different apps. Notebooks is a persistent, evolving workspace where you bundle multiple sources — documents, emails, meetings — and repeatedly ask Copilot grounded questions about them.

It's evolving fast, too. Recent updates mean Copilot Notebooks can now turn your notes into a Word document, an Excel spreadsheet or a PowerPoint deck, and draw a mind map of what's inside. That's a genuinely useful shift — instead of manually assembling a deck from your research, you point Notebooks at your sources and ask for the output format you need.

The official Notebooks getting-started guide walks through adding reference files, giving custom instructions, and sharing a notebook with your team.

What is Copilot Cowork and why does it matter?

Cowork is the biggest structural change to how Copilot works, and it reached general availability in mid-2026. Rather than a back-and-forth conversation, Cowork uses the same platform that powers Claude Cowork, designed for long-running, multi-step task automation, grounded by Work IQ so you don't need to move files and data outside of Microsoft 365.

The practical effect: Copilot Cowork brings long-running, multi-step work into Microsoft 365 Copilot, moving beyond prompts and responses toward execution that unfolds over time. You hand it a task, close your laptop, and it keeps working. One breakdown of the update put it simply: Cowork is the agentic Copilot that does longer, multi-step tasks across apps in a secure cloud-hosted environment, so work keeps running even when your device is off.

That's a genuine shift from "AI that answers" to "AI that executes" — and it's worth trying on a real task, not just a demo prompt, to see where it actually saves you time versus where you still need to check its work.

How do you decide which Copilot tool to use for a given task?

Here's a simple way to sort it:

• Quick question or draft → Copilot Chat, let Auto pick the model
• "Tell me everything about X and back it up with sources" → Researcher
• "What does this data actually mean" → Analyst
• Ongoing project with scattered files, emails, and notes → Notebooks
• A multi-step task you want handled without babysitting it → Cowork
• A repeatable workflow specific to your business → Copilot Studio, where you can create or customize agents for specific tasks or workflows, starting from a template or building from scratch using natural language

The mistake most people make is defaulting to chat for everything, because that's the interface they learned first. But chat is the blunt instrument in this toolkit now — the specialized agents exist precisely because generic Q&A isn't the right shape for research, data analysis, or long-running execution.

What should you actually try this week?

Pick one messy, real task — not a toy example. A client brief that needs summarizing against a budget. A folder of meeting notes that need turning into a project timeline. A spreadsheet of numbers that needs a second opinion. Run it through the matching tool above instead of dumping it into chat.

The gap between "I use Copilot" and "Copilot actually saves me hours" isn't about prompting harder. It's about knowing which of these five tools was built for the job sitting in front of you — and for most people, that's still the missing piece.