Building a prospect research assistant
From company research in Copilot Chat to a custom assistant with structured briefs, evaluations, and review before saving to the CRM.
Before approaching a company, a business development rep needs a useful brief: what the company does, why it might be relevant, and which facts still need checking. Gathering that information is only part of the work. Someone must review it and save it where colleagues can use it.
This series follows how we built a prospect research assistant at Aligned, from separate conversations in Copilot Chat to a custom agent with a review card and CRM access. Each iteration covers one improvement, what the tests revealed, and instructions you can use with your own team.
The problem we set out to solve
We already used Claude for company research, with people directing the work. Each run depended on someone guiding the next step, checking the output, and managing the handoffs.
We wanted fewer manual handoffs and more consistent briefs in the Microsoft tools our team already used. Microsoft 365 Copilot would be the place to request research, review the findings, and ask for a CRM save.
We focused on manufacturers and distributors. The brief needed to distinguish supported facts from possible opportunities, retain source links, and make conflicting or missing information visible.
The process
Turn company research into a useful CRM brief, with a person reviewing the findings and deciding what to save.
From research to a reviewed brief
- 01ResearchGather company facts, sources, and possible opportunities.
- 02Review and correctCheck the evidence, resolve errors, and keep unknowns visible.
- 03Save to CRMRequest the save, then check the company record and note.
What success means
- Useful: does the brief help a colleague understand the company?
- Checkable: can they follow the evidence and see what remains uncertain?
- Repeatable: does another company or a repeated request expose new problems?
- Worth the effort: how much time goes into review, correction, and recovery?
To try this with your team: research a company in a chat assistant your organization allows. Ask a colleague to check the claims and source links before you connect it to other systems.
Our tools and approach
The first prototype used Microsoft 365 Copilot Chat, with Clay as the enrichment service and Attio as the CRM. Our approach draws on Microsoft’s planning guidance to plan the work around the business task. The checks in these articles also apply when you use other tools.
Documented so far
Seven iterations cover company research, CRM saving, structured briefs, evaluations, prompt improvement, topics, and interactive review cards.
- Each includes setup instructions, screenshots, and findings. Repeated evaluations exposed remaining errors in answers and grading; CRM-save repeatability remains untested.
- The assistant supports a person who reviews the research, corrects it, and chooses what to save. Production deployment is outside this series.
Read the iterations
-
From company research to a CRM brief
Researching a company in Copilot, reviewing the findings, and saving a brief in the CRM.
Read iteration -
From general-purpose chat to a custom agent
Prompt-defined research structure and directly connected tools in one custom Copilot agent.
Read iteration -
Give the research brief a defined structure
Give the GTM Assistant a prompt tool with a saved seven-field output format for research briefs.
Read iteration -
Test whether the brief holds up
Use a small evaluation set in Copilot Studio to check evidence, corrections, and consistency before changing the prompt.
Read iteration -
Improve the brief without changing the test
Trace where evidence changes, revise the relevant instructions, and compare results using the same evaluations.
Read iteration -
Guide brief preparation with a topic
Use a Copilot Studio topic to check inputs, call the formatting prompt, and present its fields for review.
Read iteration -
Review and act on the research brief
Add clickable sources, corrections, and a CRM-save request to a research brief in Microsoft 365 Copilot.
Read iteration