An audit is one thing. Doing something with it is another. These are the prompts we feed our own audit JSON into — paired with the LLM that handles each one best.
raw_audit JSON field with every signal we surfaced. Copy that JSON and paste it where the prompt says {AUDIT_JSON}. Replace the other variables in amber boxes with your own context. Send to your model. The prompt does the rest. You ran an audit. Now you want the AI to surface which finding deserves the lead position in your cold email — not by what's interesting to you, but by what's painful to the buyer right now.
The competitor-gap field surfaced a vendor your prospect uses (or is considering). You want a 1-pager you can use on the next discovery call — prospect-specific, not generic.
You booked a discovery call. You have 18 hours. You want pre-call work done in 5 minutes, not 5 hours — and you want to be able to read the agenda mid-call without thinking.
A new lead came in. You ran an audit. Now you want one Slack message that gives the team everything they need without making them read 14 fields.
You lost a deal you thought you'd win. Run an audit on the account, then ask the AI what was hiding in plain sight that you under- or over-weighted.
A buyer at the audited company posted on LinkedIn. You want to comment in a way that's substantive (not "Great post!") and references something specific about their company without being weird.
A prospect went dark 60–180 days ago. Run a fresh audit. Use what's CHANGED about their company since then as the legitimate excuse to re-engage. Not "checking in." A real reason.
Audit-hunt for competitive intel — what a smart competitor would learn if they ran an audit on YOUR best customer or top account. Defense brief from the attacker's POV.