LoginSign Up
← Back to all sparks
O

Omni

ANALYTICS
Velocity6.3

Business intelligence and embedded analytics platform

omni.co ↗

Omni is letting AI write the semantic layer — the part BI vendors have always sold as craft.

semantic-modelingai-routinesmcpdbt-integrationembedded-analyticsaccess-control
◆Current state
Omni publishes a weekly changelog, and the AI thread runs through nearly every entry: AI Hub, visualization annotations, AI routines, model-suggestion endpoints, and now semantic model generation reaching general availability. Around that sits steady enterprise plumbing — OAuth and GitHub App authentication for database and dbt connections, access grants, embedding controls. Routines have picked up Slack as a delivery surface, and MCP configuration has moved into the product's own settings UI.
◆Where it's heading
The direction is unambiguous: Omni is pushing AI down from the question-answering layer into the modeling layer. Natural-language querying was table stakes; generating the semantic model itself goes after the labor that has historically made BI deployments slow. Running alongside, the routines plus Slack plus MCP combination points at analytics that leaves the dashboard entirely — scheduled or agent-triggered work delivered where people already work.
◆Prediction
Expect the AI model-suggestion endpoints to widen into a fuller programmatic modeling API, and routines to gain more destinations now that Slack has landed. These entries show a consistent preview-then-GA rhythm a few weeks apart, so the most recent AI features are the ones queued for promotion next.

◆Recent moves

  1. 2d ago

    AI semantic model generation hits GA

    ⚡ SPARK

    The AI thread building through months of preview features lands where it matters most for a BI tool: the semantic model itself. Paired with routines now creatable from chat, it moves Omni's AI from answering questions to authoring the layer the answers are drawn from.

    View source ↗
  2. 9d ago

    AI model-suggestion endpoints and connection OAuth

    Endpoints for AI model-suggestion generation expose programmatically what would surface a week later as generally available semantic model generation. OAuth for database connections continues the authentication cleanup running quietly underneath the AI work.

    View source ↗
  3. 16d ago

    AI routines reach Slack; MCP settings move in-app

    AI routines gain Slack as a delivery surface and MCP configuration moves into the product's settings UI. Two moves in the same direction: analytics that runs outside the dashboard, and agent connectivity that no longer requires a config file.

    View source ↗
  4. 23d ago

    AccessBoost for Apps and dbt deploy-token auth

    AccessBoost for Apps and HTTPS deploy-token authentication for dbt are enterprise plumbing rather than product direction. The dbt authentication work recurs often enough across these weeks to read as a deliberate push on that integration.

    View source ↗
  5. 1mo ago

    AI chart annotations GA; apps on by default

    AI visualization annotations reach general availability and apps become the default for new organizations — a quiet default change that says more about confidence than the feature notes do. The Notion integration extends the pattern of pulling external written context into the AI's working set.

    View source ↗
  6. 1mo ago

    Dashboard spacers, AI file uploads, embed display controls

    Dashboard spacer and divider elements alongside AI file uploads, the Notion integration, and embedding display controls. Layout polish and context-ingestion work landing in the same week is the usual texture of these releases.

    View source ↗