Punktero AI services
One place where AI assistants meet the tools and data you already work with. Connect your accounts, expose them to the assistant of your choice over the Model Context Protocol or plain REST, and run it all through a single, audited API gateway.
Authentication required for full access.
What you can do here
MCP servers
More than 40 ready-made servers for the systems you already use. Create an instance, connect your account to it, and your assistant — Claude, ChatGPT, Cursor, VS Code, Claude Code or your own agent — gets a set of tools it can call on your behalf.
REST and OpenAPI
Every server is available over plain HTTP as well, with a generated OpenAPI specification and interactive documentation. Useful for automations and clients that do not speak MCP.
Model API
One OpenAI- and Anthropic-compatible endpoint for the models we operate and the ones we buy, with per-key rate limits and a breakdown of requests, tokens and cost.
Vector stores
Managed search indexes for retrieval-augmented applications. Pick an embedding model, get an index and a key scoped to it, and query it from your own application.
Playground
Try the available models straight in the browser, including speech transcription, before wiring anything into your own code.
Access under control
OAuth, personal access tokens and tokens bound to a single server. Share what you connect with your team instead of with everyone, and see in the log who called what and when.
Some of the systems you can connect
Connecting your own accounts
Each connection is authorized by you and limited to the one service you connect. For Google services — Google Ads, Google Analytics, Google Merchant Center, Google Search Console, Google Sheets and Google Drive — you grant access on Google's own consent screen, the access covers only the instance you created, and you can withdraw it at any time by deleting the instance or in your Google Account settings. What we do with the data, how long we keep it and who it is shared with is described in our Privacy Policy.