Introducing LLM API Connections - Bring AI to Your Dashboard Without an API Key

We just released LLM API Connections, a built-in way to call an LLM from inside Squadbase. No contract with a model provider, no API key to request or store.

You can add a dashboard chat or a document search without waiting on engineering or IT to provision a key first.

Create it from a chat, not a settings screen

LLM API Connections are created from a conversation with an agent, not a settings screen. When your implementation needs an LLM, the agent proposes a connection name and a tier, and approving it makes the connection usable right away. There is no request-and-wait step for a key in between.

Every connection you create lands in the "LLM API" tab on the Connections screen. Each one carries a Squadbase LLM API and Built-in badge, so it stays easy to tell apart from a connection you set up with your own key.

Squadbase manages which model runs behind each tier

Text-generation connections use one of three tiers: Quick, Normal, or Smart. Squadbase decides which model runs behind each tier, and takes care of moving a tier to a faster or smarter model as one becomes available. The settings screen shows which model is currently behind a tier, and saving a change applies it to every dashboard already using that connection, so there is no need to pick a model separately for each one.

If you need one specific model, you can still connect with your own API key, the way every project already could. Anthropic, Gemini, and OpenAI are all supported, and a project already connected with its own API key keeps working as-is.

Where teams are already using it

Search over internal documents

Turn policy documents, manuals, or a maintenance ledger into a RAG search that returns an answer with a source page attached.

A chat next to the dashboard

Add a chat beside a sales or usage chart so a question that comes up while looking at it, like a breakdown by channel or a comparison with last week, gets answered on the spot instead of sending anyone to a different screen. The same connection can answer questions that span more than one dashboard. Upgrading the model behind it later changes nothing about where the chat sits or how it's called.

Start Today, No API Key Required

Any other place your project needs to call an LLM works the same way. Read the docs for the full setup.

Every update, including this one, is on the Squadbase changelog.

Squadbase is free to start, no credit card required.