Early access · Limited spots

Claude Code ships it.
Vigil watches it.

One SDK. One dashboard. Behavioral PMF score, AI cost per feature, and autonomous incident response — without PostHog, Grafana, Sentry, or a platform team.

✓ You're on the list. We'll reach out before public launch.

No spam. Early access invite when we open the beta.

Instrument in 60 seconds

# 1. Install pip install vigil-sdk # 2. Add one decorator to your AI function from vigil import trace, init init("your-api-key") @trace async def generate_response(user_id, prompt): response = await openai.chat.completions.create(...) return response # That's it. Vigil now tracks: latency, token cost, # error rate, and user behavior — automatically.

First data appears in your dashboard within 60 seconds. No config files. No alert tuning. No platform team.

📈

Behavioral PMF Score

Computed from retention curves and aha-moment detection — no survey required. The first tool that measures PMF from your actual user behavior.

First of its kind

💸

AI Cost per Feature

Know exactly which features are profitable and which are burning your OpenAI budget. Per-user, per-model, per-endpoint — in real time, not on your monthly invoice.

⚡

Autonomous Incident Response

A policy engine watches your production 24/7 and acts when something fires — auto-rollback, traffic shift, GitHub PR — while you sleep. Full audit trail of every action.

Sentinel

🎯

Three Personas, One Platform

Founder view: growth + PMF. On-call view: incidents + health. AI view: quality + cost. Same data, surfaced differently for what you need right now.

"We build AI SRE for Fortune 500 companies. What's ironic is that while building Traversal, we didn't have a tool like Vigil for our own stack. Every early-stage AI team needs this."

Raaz Dwivedi · Founder, Traversal — $48M, Sequoia + Kleiner Perkins

Claude Code ships it.Vigil watches it.