Decision-heavy assistants that turn operational signals into short, reviewable recommendations.
Agents interpret and recommend. n8n performs the scheduled collection, sync, and routing around them.
Turn substantive GHL sales calls into aggregate buyer-demand, friction, and momentum signals.
Qualified GHL call transcripts · Opportunity and contact context
Supabase communication_insights · Aggregate call_intel_reports
Monitor priority competitors by market and summarize useful movement into the Intel Pipeline.
DB: IPA Competitors · Competitor websites and public search results
DB: IPA Intel Intake summaries · Supabase intel_briefs and sync_runs
Interpret detected source-shape changes, explain likely impact, and prepare approval-gated repair handoffs.
Supabase schema_drift_events · Supabase schema_watch_snapshots
Slack digest and review thread · Recommended action classification
Answer staff questions from approved Supabase-backed operational data with source links and explicit caveats.
Supabase dashboard data · Published intelligence and approved resources
Source-cited IPA Hub answers · Supabase agent_runs usage metadata
Synthesize sales, marketing, operations, pipeline, content, and intel signals into leadership-ready briefs.
GHL and marketing data mirrored in Supabase · IPA ClickUp operational data
Source-traceable leadership brief · Structured executive report
Synthesize automated and team-submitted intel into short, reviewed, staff-facing briefs.
DB: IPA Intel Intake · DB: IPA Team Intel
Draft DB: IPA Intel Briefs records · Sales talking points
Turn scheduled RSS, web, social, and newsletter collection into compact, traceable intake summaries.
Approved external channels · Market and topic priorities
Summarized Intel Intake records · Source links and relevance caveats
Registry separates real from draft: agents/registry.ts is the canonical status registry; specifications do not become active merely because a file exists.
Agent means judgment: Use agent language for synthesis, diagnosis, recommendations, triage, drafting, coaching, or QA.
n8n owns glue: Use workflow automation for schedules, API calls, syncs, retries, alerts, and record movement.
Human review stays visible: Agent outputs should show review status before staff-facing publication or production repair.
Turn approved intel and evergreen context into structured content ideas and briefs.
Published Intel Briefs · Knowledge Vault context
Content ideas · Structured channel-specific briefs
Create first-pass content from approved briefs and source context.
Approved content brief · Brand, source, and channel rules
Draft Content Pipeline records
Check drafts against source accuracy, brand rules, compliance requirements, and operating procedures.
Content draft and source brief · Operational Documents and brand rules
QA notes · Required edits and source warnings
Identify reviewed intel that has become durable, reusable business knowledge.
Intel Briefs · Team Intel and competitor summaries
Suggested Knowledge Vault additions or updates
Diagnose pipeline, conversation, follow-up, objection, and closing patterns for sales leadership.
GHL pipeline and activity history · Conversation summaries and sales SOPs
Performance insights · Risk flags and coaching recommendations
Recommend next-best actions, objection handling, timing, and information needs for active opportunities.
Active opportunity context · Conversation summaries and approved sales guidance
Deal strategy notes · Human-reviewed follow-up suggestions
Analyze channel, campaign, content, lead-quality, and conversion performance and recommend improvements.
Marketing and analytics mirrors · GHL attribution and outcomes
Performance summaries · Campaign, content, and landing-page recommendations
Identify recurring agent failures, stale prompts, missing context, and automation reliability gaps.
Supabase agent_runs and sync_runs · n8n execution notes and human review outcomes
Prompt and runbook suggestions · Reliability and evaluation gaps