A Company Brief Under Five Credits
Outcome: a briefing on any company — what they do, what they run, how big they are, what they are building — in one prompt, for under three credits.
- Surface
- MCP server
- Level
- Beginner
- Uses
- enrich_linkedin_page · find_company_techstack · find_company_website_traffic · company_job_listings
- Credits
- ~2.8 per company
- Prerequisite
- Lesson 02
What each tool contributes
| Tool | Answers |
|---|---|
| enrich_linkedin_page | Who they are — size, industry, location, description |
| find_company_techstack | What they run — and whether a competitor is in it |
| find_company_website_traffic | How big they really are, and which way they are going |
| company_job_listings | What they are building right now — the timing signal |
The last one is the most useful per credit. A job post names internal tools, team size and the problem being solved — none of which the other three can tell you.
The prompt
Using the SyncGTM MCP server, brief me on acme.com before a call.
Steps:
1. enrich_linkedin_page for the company.
2. find_company_techstack.
3. find_company_website_traffic.
4. company_job_listings, last 30 days only.
Then summarize in this format:
WHO THEY ARE: one sentence
SIZE: headcount and traffic band
STACK: any tools relevant to sales, marketing or data — list them
HIRING: roles posted in the last 30 days, grouped by function
ONE THING WORTH ASKING: a single question based on the above
Do not call any other tools.The format block is what turns four tool outputs into something you can read in twenty seconds before a call, rather than four blocks of JSON.
“One thing worth asking” is the highest-value line in the brief. It forces the assistant to synthesize rather than list, and it gives you an opening question that is specific to this company — which is the entire point of doing research before a call.
Read it in this order
Hiring first
It is the only part with a date on it. Recent roles tell you what is happening now, which is what a conversation should be about.
Stack second
A detected competitor or complement changes the whole conversation. A positive detection is strong evidence; a negative one means nothing.
Size third
Context for everything else, and it tells you which of your case studies is relevant.
Description last
It is the least surprising part and the most likely to be marketing copy.
Keeping it cheap
| Habit | Saving |
|---|---|
| Cap job listings to 30 days | Avoids pulling a long history you will not read |
| Skip traffic for B2B SaaS where headcount is reliable | ~0.5 credits |
| Skip stack where your product does not displace anything visible | ~0.5 credits |
| Never ask for “everything you can find” | Unbounded |
Two tools — page enrichment and job listings — get you most of the value for about a credit. Add the other two only when they change what you would say.
For several companies
Do this for these 10 domains: [...]
Cap at 10. Return one row per company with the same fields.
If a tool returns nothing for a company, leave that field blank and continue.Ten companies at under three credits each is a morning of manual research replaced by one prompt.
Do this now
Run it on a company you know well
The only way to judge accuracy.
Check the stack against reality
Do the detections match what you know they use?
Read the “one thing worth asking”
Is it a question you would actually ask? If not, the format needs tightening.
Run the two-tool cheap version
Page enrichment plus job listings. Compare what you lost.
Run it for ten companies
With the cap and continue-on-failure lines.
Note the credits
Against the ~2.8 estimate.
Check your work
- The output follows your format, not the tools’ raw shapes
- Hiring data is capped to a recent window
- You checked the stack detection against a company you know
- The suggested question is specific to the company
- Credit use matched the estimate
Where this breaks
Stack detection reads what a website exposes publicly, so a negative result means “not detected”, not “not used”. Walking into a call believing a company does not use a competitor’s product because the brief did not list it is a bad surprise waiting to happen. Treat positives as evidence and negatives as silence.
Further automation
Point this at your calendar rather than at a domain, and it briefs every meeting before you join. That build is AI-Powered GTM lesson 18, and it is the same chain with a different input.
Next lesson
04 — Backfill profiles from an email list, turning bare addresses into records.
Reference for this lesson: enrich_linkedin_page, find_company_techstack, company_job_listings, MCP prompting guide.