Run a multi-source competitive teardown with the Researcher agent
Give Copilot's Researcher agent a question and get back a structured, source-cited report drawn from multiple documents.
Give Copilot Cowork a detailed brief for a report drawing on months of data and get back a completed deliverable to review, not just a draft outline.
A completed executive-ready report produced without a person assembling it from scratch, freeing that time for review and judgment calls instead of compilation.
The apps run in this order.
Spell out the report's audience, required sections, the specific data sources to draw from, the time period covered, and any formatting standard (house style, required charts) — Cowork executes the brief you give it, so vague instructions produce a vague report.
Give it access to the actual underlying files and locations (the quarterly financials workbook, the project tracker, last quarter's report as a style reference) rather than summarized inputs, since the value here is it doing the synthesis work itself.
Prompt idea:
Build a Q3 board report covering revenue performance by region, the top 3 project risks from the program tracker, and headcount changes. Use the same structure as last quarter's report, pull numbers from the finance workbook and project tracker, and flag anything you can't find a source for instead of estimating.
Unlike a quick prompt-and-check cycle, this is meant to run unattended on a substantial task — review the completed report as a whole deliverable, checking structure and figures rather than babysitting the process.
Treat the output as a strong first draft from a new analyst, not a finished, trusted report — every number and claim needs verification before it goes in front of leadership.
Source: Microsoft, "Copilot Cowork is now generally available" (June 2026)
A complete draft report — structured, sourced, and formatted — ready for an executive review pass rather than a first-draft outline.
Pulling revenue performance, top project risks, and headcount changes from several separate source systems into one coherent report is exactly the large, multi-source synthesis job worth delegating.
Give Copilot's Researcher agent a question and get back a structured, source-cited report drawn from multiple documents.
Use a Case and Precedent Analysis Agent to process a large volume of legal documents, surface usable arguments, and draft a first-pass brief.
It's end of quarter and the raw numbers just came in.
Workflows like this tend to raise real governance and licensing questions once more than one person is using them — that's exactly what we help with.