The tools we use and the guides for running things. Get access to each from a founder (most via 1Password).
Tools
| Tool | What we use it for |
|---|---|
| GitHub | Code, PRs, CI |
| Linear | Issues, specs, cycles |
| Slack / Discord | Internal comms / community |
| GCP | Hosting (Honcho tenants, Groudon, Xatu) — GKE, AlloyDB, Cloud Run, Vertex |
| ArgoCD | Deploys everything on GKE from git |
| Fly.io | Mostly unused now, has legacy grafana dashboards and litellm proxy |
| Vercel | Frontend hosting |
| Supabase | Dashboard auth (JWT, MFA) |
| AWS | Secrets Manager (instance secrets), S3 (Parquet archive) |
| Redpanda | Kafka for the Xatu billing pipeline |
| Grafana + Mimir | Metrics and dashboards |
| Tailscale | Access to internal services and the GPU boxes |
| Langfuse | LLM tracing |
| Sentry | Error tracking |
| CodeRabbit | Automated PR review |
| Cloudflare | DNS, Pages, Workers |
| Modal | GPU, Inference |
| PostHog | Product Analytics |
| Rippling | HR / payroll |
| Azure | Devboxes |
| Rootly | ON-call and incident management |
| Arcjet | Web Application Firewall in code |
| Granola | Meeting Recordings & Transcripts |
| Superhuman | Email tracking and scheduling (not everyone) |
Guides
- Where to look — when you want to know X, which console, and how to reach it.
- Platform knowledge transfer — how billing, observability, caching, secrets and CI actually work, and what to check when each misbehaves.
- Internal apps — the apps we host for ourselves (PR dashboard, Falinks) and how to get in.
- Code standards — the toolchain, the canonical lint configs, and how we gate them.
- Incident Response & On-Call — Rootly topology, status page, on-call rotation.
- Google Cloud support — the Slack channel with our reps, and how to file a case.
- Load testing — how we load-test the platform.
- DGX (Varoom) agent setup — notes on the DGX box.
- Plastic Labs agent plugin — our shared skills and MCP servers, in one install.
To understand how a change reaches production, read the GitOps pipeline. Nothing on GKE is applied by hand.
For operational procedures (release, debugging), see Runbooks.