What problem does it solve? Inbound and sent email contains client conversations, deal signals, deadlines, and attachments that never make it into the CRM, forcing teams to manually log activity and risking lost follow-ups on consulting deals. ## Core Features & Use Cases - Body-first email classification: Uses Claude Sonnet to read each email body and decide which consulting clients, deals, and listings are discussed, with participant matching as a secondary signal. - Automated CRM writes: Creates and updates Airtable records (Consulting Clients, Deals, Contacts, Tasks, Activity, Listing Notes) including auto-stage moves, deal value updates, and task creation from stated deadlines. - Slack notifications and review queue: Pings each client's owner with detailed write summaries, queues overnight digests, and flags ambiguous threads as CRM_NEEDS_REVIEW for daily audit. - Use Case: A team member forwards a client thread about a hotel deal; the cron extracts the referenced companies, appends a continuation note to the existing thread's activity rows, creates a task for the stated deadline, and pings the deal owner in Slack within five minutes. ## Quick Start Ask Claude Code to review or modify the email-to-CRM classification rules in this skill and mirror the changes in api/cron/process-emails.js.
