Business Meta-Analysis

Collect signals from YouTube, CRM, cron jobs, social, email, Slack, Fathom meetings, and HubSpot — then compact them into the top 200 signals by confidence.

Inspired by Brian Armstrong (Coinbase CEO) using AI to review all business data. Collect every signal: YouTube metrics, CRM health, cron reliability, social growth, Slack messages, emails, Asana tasks, X/Twitter, Fathom meeting transcripts, and HubSpot pipeline. Compact everything down to the top 200 signals ranked by confidence. This becomes the input for your AI council.

Have a growth strategist, revenue guardian, skeptical operator, and team dynamics architect all review and debate the compacted signals before generating recommendations.

Phase 1: A draft analysis reviews all 200 signals. Phase 2: Four specialized agents — growth strategist, revenue guardian, skeptical operator, team dynamics architect — review and debate the draft. They collaborate, push back on each other, and reach consensus. A council moderator (Opus) reconciles disagreements, ranks everything, and produces the final report.

Run your nightly business briefing during sleeping hours when you're not competing for API quota or generating interactive requests.

The full business meta-analysis is expensive — it uses Opus for the council moderator and runs multiple agents. Schedule it to run in the middle of the night when you're sleeping and not making interactive requests. This avoids competing for API rate limits and lets the heavy computation happen when resources are cheapest.

Persist all data you possibly can in local databases. Historical data becomes exponentially more valuable as patterns emerge over weeks and months.

Contacts, knowledge base entries, video pitches, business analyses, social media snapshots, cron logs — store it all. Use the hybrid SQL + vector pattern for every data type. Historical data lets you spot trends, measure improvement, and give your AI council better context. The nightly briefing gets smarter every day because it has more historical data to reference.