Features

What Omni can do
across your company

Omni can carry out tasks, work with company context, use the models you choose, and operate within your existing permissions.

Get early access View on GitHub
For employees

Work across company tools from one conversation

Omni can gather context, analyze information, and carry out actions across connected systems.

1

Carry out work across tools

Omni can retrieve information and take actions through the same connected systems employees already use.

2

Answer with company context

Responses draw from permission-aware company information and include the sources behind the answer.

3

Analyze files and data

Omni can use isolated Python and shell environments for tasks that require calculation, transformation, or deeper analysis.

Claude Sonnet
Prepare this week’s launch update, identify the biggest blocker, and share it with the product team.
SlackJiraDocsDrive
searched Slack, Jira, Docs, Drive: launch progress and blockers
fetched: Product launch plan
ran: Python launch readiness analysis
completed: Slack > send update to #product

I prepared the launch update, identified the mobile release as the main blocker, and posted the summary with owners and next steps to #product.

Ask a follow-up...
Claude Sonnet

Company context that respects source permissions

Connectors keep information current, hybrid search retrieves it, and source permissions determine what each employee can use.

Google Drive
Slack
Jira
Confluence

Maintained workplace connectors

Sync company information and expose supported actions from Google Workspace, Slack, Jira, Confluence, and other workplace systems.

✓ Product launch plan Available
– Compensation review Restricted

Permission inheritance

Source permissions are carried into Omni so retrieved context stays scoped to the employee making the request.

BM25 + Semantic → Ranked context

Hybrid search and retrieval

ParadeDB BM25 and pgvector semantic search work together to retrieve relevant company context from Postgres.

Jira LAUNCH-42 · Mobile release blocker ↗
Slack #product · Launch status thread ↗

Open the original source behind an answer

Source citations you can verify

See which documents, messages, and tickets Omni used, then open the original source to check the evidence behind an answer.

Control models, access, and usage

Choose the models Omni can use, manage employee access, audit activity, and see where model spend comes from.

Anthropic
OpenAI
Gemini
Azure
Bedrock · Vertex AI · OpenAI-compatible endpoints

Use the models you choose

Connect commercial APIs, managed cloud models, or self-hosted open-weight models without rebuilding the company context layer.

Anthropic usage and spend
Gemini usage and spend
Self-hosted usage tracked

Cost and token visibility

Track token usage and spend by team, model, and request, then compare commercial and self-hosted model usage from one place.

Explore cost management →
✓ SSO and SCIM
✓ Audit logs and exports
✓ Model and access policies

Governance and administration

Manage employee access, set model policies, and keep a record of queries, model calls, source access, and administrative changes.

Deploy and extend Omni

Run the open-source system in your environment and add the integrations your company needs.

01

Open-source core

Apache 2.0 licensed and inspectable from ingestion through agent execution.

02

Docker and Terraform

Use Docker Compose or production Terraform modules for AWS and GCP.

03

Native MCP support

Connect MCP servers so Omni can use the tools they expose.

04

Connector SDKs

Build company-specific connectors in Python or TypeScript.

05

Postgres-powered retrieval

ParadeDB BM25 and pgvector semantic search in the same data layer.

06

Managed or self-hosted

Operate Omni in your environment or use Omni Cloud when you want us to run it.

See how Omni works →

Put Omni to work across your company

Join early access, explore the open-source repository, or read the deployment and integration guides.

View on GitHub ↗ Read the docs ↗