Aperture: Your Team's Private AI Gateway

My notes

Summary

Aperture is a Tailscale-native proxy layer that centralises AI provider access (OpenAI, Anthropic, Google, etc.) across a team’s tailnet, eliminating the need to distribute API keys to individual machines. It is currently in alpha and free during the preview period. The pitch is frictionless governance: one gateway, one set of credentials, full visibility.

Key Insight

• Core problem solved: API key sprawl, every developer having their own key creates audit, rotation, and cost-attribution headaches. Aperture pushes this to a single choke point inside the private network.
• Tailscale-native advantage: because traffic stays on the tailnet, there is no public endpoint to secure or firewall separately. Access control is inherited from the existing WireGuard mesh.
• Alpha / free-tier timing: early adopters can test without cost commitment. Worth trialling now before pricing is set, especially for teams already running Tailscale.
• Governance angle: centralised gateway enables usage logging, per-team quotas, and model routing rules in one place, capabilities that are otherwise scattered across provider dashboards or homegrown middleware.
• Limitation: content is sparse (alpha landing page); actual configuration depth, supported models, and audit log granularity are unknown until tested.