# AI Lab Outreach System

> Systematic 4-week outreach system for AI labs (Anthropic, OpenAI, Google, Meta, Cognition, Cursor, and 13 others) to sell Bouts data licensing and private benchmarks — from warm-up through commercial conversation. Use when prospecting AI labs, writing outreach copy, or building the pipeline for data licensing revenue.

- Skill: `nickgallick/ai-lab-outreach-system` (Agent Skill)
- Install (CLI): `npx skillmds add nickgallick/ai-lab-outreach-system`
- Raw SKILL.md: https://api.skillmd.com/api/skills/nickgallick/ai-lab-outreach-system/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: DevOps & Infra
- Author: nickgallick (https://skillmd.com/u/nickgallick)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/nickgallick/ai-lab-outreach-system

---


# AI Lab Outreach System

## Target list

### Tier 1 — Frontier model labs
- Anthropic (Claude)
- OpenAI (GPT)
- Google DeepMind (Gemini)
- Meta AI (Llama)

### Tier 2 — Agent-focused companies
- Cognition (Devin)
- Sourcegraph (Cody)
- Cursor
- Replit
- Poolside AI

### Tier 3 — Enterprise AI
- Microsoft (Copilot)
- Amazon (CodeWhisperer/Q)
- IBM (WatsonX)

## 4-week outreach sequence per lab

### Week 1 — Warm-up
- Follow their AI researchers and DevRel people on X
- Engage genuinely with 3-5 posts (technical comments, not "great post!")
- Share Bouts data that mentions their model positively if applicable

### Week 2 — Soft touch
Send DM or email to DevRel team with ONE specific data point.

"We ran 200 agents on our competitive benchmark this month. [Model] showed interesting strength on Recovery — scored 15% above average. Thought your team might find the full breakdown useful."

Attach: one-page performance summary of their model (free sample).

### Week 3 — Value offer
"We're publishing our monthly AI Agent Index next week. We can include a dedicated section on [Model] performance with detailed failure archetype analysis. Would your team find that valuable?"

This positions the data as something they WANT, not something we're selling.

### Week 4 — Commercial conversation
"Several labs have asked about private benchmark lanes — running pre-release models against our challenge suite before public launch. Would [Lab] be interested in exploring that?"

Now the conversation is about THEIR need, not our sale.

## Never do
- Cold pitch the paid product first
- Send generic "partnership" emails
- Claim Bouts is the "best" or "only" benchmark
- Share negative data about their model in initial outreach

## Tracking
Build a simple CRM row per target:
- Lab name
- Primary contact (name, role, LinkedIn, X)
- Current week in sequence
- Last message sent
- Response received
- Next action