Does Blits.ai support MCP?

Yes, in both directions. Agents call tools on MCP servers you connect, and the platform hosts an MCP server per app that gives other agents tools over the data you store in the platform.

Does Blits.ai support the A2A protocol?

Yes, in both directions. Blits.ai agents are published with an Agent Card for other platforms to call, and external A2A agents, for example on Salesforce or ServiceNow, connect as tools for Blits.ai agents.

Can we use our own model or API key?

Yes. Any model behind an OpenAI compatible endpoint works with your own key, set per agent, including models you host yourself.

Which model providers does Blits.ai support?

More than 300 models, from OpenAI, Anthropic, Google, Mistral, DeepSeek, Meta, Alibaba, Humain and Fanar to OpenAI's open weight models, plus your own OpenAI compatible endpoint.

Yes, REST and WebSocket APIs to start conversations, send messages, stream answers and run workflows, documented on the API documentation page.

Integration standards

Blits.ai connects to your systems, models and other agents through open standards where they exist: MCP and A2A in both directions, REST and WebSocket APIs, BPMN for flows, OpenTelemetry for logs and OpenAI compatible endpoints for your own models.

Each capability is marked as on by default or configurable per assistant, workflow or tenant. Defaults that touch customer data can be switched off where regulation or your own policy requires it.

Open standards

The protocols the platform speaks, as a client and as a server.

MCP client

Configurable

Agents call tools on any Model Context Protocol server you connect. Servers are health checked every hour, and a failed connection is retried automatically.

MCP server

Configurable

The platform hosts an MCP server per app, protected by an API key, that gives other agents and assistants tools over the data you store in the platform.

A2A (Agent2Agent)

Configurable

Blits.ai agents are published with an Agent Card, so agents on other platforms can discover them and hand them tasks with streaming status updates. External A2A agents connect as tools for Blits.ai agents.

REST and WebSocket APIs

Start conversations, send messages and stream answers over REST or WebSocket, and run workflows through the API. The full reference is in the API documentation.

BPMN flows

Conversation flows are modelled in BPMN, the open standard for business processes, so they can be read and reviewed outside the platform.

OpenTelemetry

Configurable

The decision log of every conversation streams over OpenTelemetry (OTLP) to your own SIEM or observability stack, or exports as JSON.

OpenAI compatible endpoints

Configurable

Any model behind an OpenAI compatible endpoint can be used with your own key, including models you host yourself. Speech recognition and voice have their own custom endpoints too.

Model providers

Choose a model per assistant and per task, and switch without rebuilding.

See every supported model and speech service

300+ models

More than 300 models from OpenAI, Anthropic, Google, Mistral, DeepSeek, Meta, Alibaba, Humain, Fanar and OpenAI's open weight models, plus your own endpoint with your own key.

Bring your own key

Configurable

Set your own API key and endpoint per agent. Usage is tagged as platform key or customer key, so costs stay traceable.

Business systems

Prebuilt connectors for common systems, and functions for everything else.

See all integrations

Prebuilt connectors

Configurable

Microsoft 365, SharePoint, OneDrive, Microsoft Teams, Dynamics 365, Power BI, HubSpot, Freshdesk, Zoho, Odoo, Notion and Moneybird, each with built in tools agents can call.