For CTOs and CIOs · Platform and OEM teams

Embed enterprise search and RAG in your product, without moving anyone's data.

Your customers' knowledge lives in Microsoft 365, SharePoint, Salesforce, file shares, and databases you will never be allowed to copy. SWIRL is the knowledge layer you embed instead: it federates the search, elects the one canonical document among the copies, and grounds the AI answer, all inside the tenant where the data already lives.

60 seconds: sign in, federate a search across Microsoft 365, watch SWIRL elect and pin the canonical document, then get a grounded answer that cites only that document.

Why embed SWIRL

Three things your platform gains.

Reach

Expand your front-end to every customer silo, without moving data.

SWIRL queries the customer's systems live, in place: Microsoft 365, SharePoint, Salesforce, iManage, Box, Snowflake, databases, and 150+ sources. No connector-by-connector ETL project, no copy of the customer's corpus to secure, and their permissions (SSO, OAuth2, OIDC) are enforced on every query.

Control

Use internal, private LLMs to access and summarize it.

SWIRL works with any LLM: the customer's Azure OpenAI or Anthropic tenancy, or a fully on-prem model served by Ollama. Regulated customers keep inference inside their walls. You are never locked to one model vendor, and neither are they.

Experience

Deliver the ChatGPT experience, centered on their data.

Search, AI summaries, and a conversational assistant, all grounded in the customer's own documents and cited back to them. SWIRL's canonical election means the answer comes from the current version of a document, not whichever of nine stale copies scored highest.

How you integrate

Three integration paths, one knowledge layer.

SWIRL deploys as a service in your tenant or your customer's: Docker Compose, Kubernetes, or a cloud marketplace image. Your product talks to it three ways.

REST API

Every capability in the demo above is an endpoint: run a federated search, read ranked results, generate a grounded summary, pin a canonical document. Build your own UI on top; SWIRL is the backend.

MCP server

SWIRL ships an MCP server, so agents you build (or Claude, Copilot, and other MCP clients your customers already use) can search, read documents, and get grounded answers as native tools.

White-label UI

The Galaxy interface in the video ships with SWIRL and takes your branding: logo, colors, and domain. Fastest route to shipping search and an assistant under your own name.

150+ connectors
M365, SharePoint, Salesforce, iManage, Box, Snowflake, SQL, and more
Data stays in place
Live federated queries; the corpus is never copied or re-indexed wholesale
Permissions passthrough
SSO, OAuth2, and OIDC enforced per user, per query
Any LLM
OpenAI, Anthropic, Azure, Google, or on-prem via Ollama

Put it in front of your architects.

A 30-minute technical session with a SWIRL engineer: your stack, your customers' sources, and what embedding actually takes. No slides you have already seen; we run it live.