The GTM operating system for AI-first sales orgs. Spec + scaffold + skills + agents + reporting, built on Salesforce + n8n + Claude.
A complete operational reference for running go-to-market in autopilot mode — humans-in-the-loop only for approval and strategy. One of a series of *-ops repos that automate functional domains.
| Role | What you'll find here |
|---|---|
| 🛠️ GTM Operators (RevOps, Demand Gen, Marketing Ops) | The 7 agents that run the engine + the 3 BI surfaces that report on it |
| ⚙️ GTM Engineers | Reference architecture for an agent layer that sits above any CRM |
| 📣 PMMs | Lifecycle + cross-sell + churn-save motion playbooks |
| 👔 CMOs / VPs | Operating-model spec, comp implications, org structure (HoDG / GTM Engineer roles) |
| 🚀 Indie builders | Reference architecture for a productized "Developer Growth OS" |
┌─────────────────────────────────────────┐
│ ACQUISITION LOOP │
│ ICP-Scout · Outreach-Writer · Reply │
└────────────────┬────────────────────────┘
│
┌────────────────▼────────────────────────┐
│ NURTURE LOOP │
│ Stage-Mover · Cross-Sell-Detector │
└────────────────┬────────────────────────┘
│
┌────────────────▼────────────────────────┐
│ RE-ENGAGEMENT LOOP │
│ Dormant-Detector · Churn-Saver │
└─────────────────────────────────────────┘
All running on: Salesforce + n8n + Claude Max +
Drive AI + Google Forms/Sheets +
Smartlead (20 domains) + MoEngage +
Clay + ZoomInfo + Sales Nav +
Ahrefs + SimilarWeb +
Metabase + AWS QuickSight
13 reliability rules · DIN approval gate · UTM tagging mandatory · 3-surface BI rule (anti-sprawl) · mart_buyer_journey as the spine of all reporting.
| You are... | Start here |
|---|---|
| New here, want the 60-second tour | 👉 OPERATOR-QUICKSTART.md |
| Hands-on: 7-day step-by-step walkthrough | 🛠️ TUTORIAL.md — clone → seed → query → first DIN → first weekly digest |
| Building it for your company | 📘 docs/gtm-context.md (the full ~1,400-line spec) |
| Want the architecture diagram | 🏗️ docs/architecture.md |
| Extending with new agents/skills | 🤝 CONTRIBUTING.md |
| What's planned for v1.5 / v2 | 🗺️ ROADMAP.md |
| Version history | 📋 CHANGELOG.md |
| Component | Status | Notes |
|---|---|---|
Spec (docs/gtm-context.md) |
✅ Stable v0.1 | Industry-validated against CS2 / Domestique / Factors / Clay |
Python reference scaffold (src/) |
🟡 Partial | 2 of 7 flows fully scaffolded (LangGraph). Use as a structural reference, not production code. |
Claude Code skills (skills/) |
🔲 Not yet built | 11 skills planned per spec §5 |
n8n agents (agents/) |
🔲 Not yet exported | 7 + 3 utility workflows planned |
dbt-lite marts (sql/) |
🟡 Schema only | mart_buyer_journey (the spine) is P0 to build first |
Dashboard templates (dashboards/) |
🔲 Not yet built | 12 Sheets · Metabase JSON · QuickSight definitions |
| Production deployment | 🔲 None | This is a private design repo; production lives inside mothi systems |
gtm-ops/
├── README.md ← you are here
├── OPERATOR-QUICKSTART.md ← 60-second tour for new operators
├── CONTRIBUTING.md ← how to add a skill / agent / mart
├── ROADMAP.md ← v1 / v1.5 / v2 phasing
├── CHANGELOG.md ← version history
├── LICENSE.md
│
├── docs/
│ ├── README.md ← documentation TOC
│ ├── gtm-context.md ← canonical mothi GTM spec (~1,400 lines)
│ ├── architecture.md ← 3-layer architecture detail
│ └── internal/ ← confidential narrative (session log, pitch, demo script)
│
├── src/gtm_ops/ ← Python LangGraph reference implementation
│ ├── flows/ ← 7 agent flows (2 scaffolded, 5 stubbed)
│ ├── integrations/ ← 6 typed clients (HubSpot · OpenRouter · Fathom · Smartlead · Dripify · Typefully)
│ ├── cli.py ← `uv run gtm-ops <command>`
│ ├── config.py ← typed settings
│ └── models.py ← Pydantic schemas
│
├── skills/ ← Claude Code skills (markdown-first)
├── agents/ ← n8n workflow JSON exports
├── sql/ ← dbt-lite marts (Postgres views)
├── dashboards/ ← Sheets / Metabase / QuickSight templates
├── evals/ ← Promptfoo prompt regression tests
└── notebooks/ ← one-off analysis (forecasting, win/loss clustering)
This repo sits inside a broader personal-OS pattern:
| Repo | Domain | Status |
|---|---|---|
gtm-ops |
Go-to-market operations (this repo) | spec stable, build pending |
mothi-os (private) |
Master personal OS — 23 active skills | live |
llm-wiki (github.com/mothivenkatesh/llm-wiki) |
Durable knowledge layer — entities, concepts, sources | live; cross-linked from gtm-context |
Future: pmm-ops, partner-ops, content-ops, community-ops |
Per-domain operating systems | planned |
Vision: every functional domain runs in autopilot — humans-in-the-loop only for approval (DIN) and strategy. Each *-ops repo follows the same shape: spec markdown + Claude Code skills + workflow exports + SQL marts + dashboards.
# 1. Copy env
cp .env.example .env # fill in API keys
# 2. Spin up Postgres + Metabase + Redis
docker compose up -d
# 3. Install Python deps (uv recommended)
uv sync
# 4. Seed schema
make seed
# 5. Run the canonical flow
uv run gtm-ops meeting-prep --deal-id=1
# 6. Run evals
cd evals && npx promptfoo evalNote: the Python scaffold is a reference implementation. The production system runs on n8n + Claude Max API + the spec — not this Python code. Use this scaffold to understand the shape; build production in n8n.
Private / proprietary. See LICENSE.md. Do not distribute, fork, or share without explicit permission.