Analytics
One governed definition of every metric. Every way to ask about it.
Kontala Analytics is an agentic BI platform built on a governed semantic layer. Dashboards, self-serve exploration, AI agents, embeds and APIs all return the same numbers.
Six reasons teams trust their numbers again
One source of truth for metrics
“Revenue” is defined once, in code, and reused by dashboards, AI, embeds and APIs.
Ends “whose number is right?” meetings and cuts reconciliation hours.
Trustworthy AI on governed data
Agents answer from curated metrics, show the query they ran, respect each user's permissions, and say so when data is not available.
Self-serve without a SQL-fluent intermediary, and without the risk of “chat with your database”.
Self-serve without the chaos
Business users drag, filter and pivot pre-built metrics and dimensions in a point-and-click UI.
The data team stops being a ticket queue and time-to-answer drops from days to minutes.
Built for the AI-coding era
Claude Code, Cursor and Codex can build dashboards via MCP and installable agent skills, with humans reviewing in a PR.
Data teams produce more with the same headcount.
Enterprise security and governance
Row-level security, custom roles, SSO/SAML/SCIM, audit logging and service accounts.
Passes security review and is safe to expose to customers and partners.
Fast on the warehouse you already have
Queries run in your warehouse. Caching and pre-aggregates serve busy dashboards without repeated heavy queries.
Lower warehouse spend, snappier dashboards, and no new data silo.
Everything on one governed layer
Semantic layer
- Metrics, dimensions, joins and descriptions defined in YAML, on dbt or standalone.
- Rich metric types, parameters, virtual views and custom fields.
- Metrics catalog with categories, owners and a Metrics Explorer.
- Pre-aggregates served from stored results instead of the warehouse. Beta · Enterprise
- Metrics SQL API over the Postgres wire protocol. Beta · Enterprise
Exploration and analysis
- Point-and-click Explorer that shows the generated SQL for every result.
- Table calculations, custom dimensions and metrics, conditional formatting.
- SQL Runner, funnel builder, date zoom and period-over-period comparisons.
- Upload CSVs or connect Google Sheets and query them alongside warehouse data.
- Merge queries from two explores on a shared key. Beta
Visualisation
- Bar, line, area, scatter, pie, funnel, treemap, gauge, maps, Sankey, tables and pivots.
- Custom charts with Vega-Lite and a chart library.
- Dashboards with tabs, filters, parameters and tile comments.
- Version history with side-by-side comparison, restore and recently deleted recovery.
- Documents for narrative pages, plus brand theming.
AI and agents
- Ask AI in the app, in Slack and in Microsoft Teams, pinned to the chart or dashboard you are viewing.
- Governed answers with the underlying query inspectable and per-user permissions carried through.
- Verified answers, agent memory, knowledge documents and specialised agents with their own data scope.
- Evaluation suites and review items routed to Linear or Jira.
- MCP server to connect Claude, ChatGPT, Codex, Cursor and VS Code over OAuth.
- Deep research, AI writeback and Autopilot. Beta
Delivery and collaboration
- Scheduled deliveries to email, Slack, Teams and Google Chat as CSV, XLSX, image, PDF or Google Sheets.
- Threshold alerts on any metric.
- Shareable links, comments, review requests and mobile push notifications.
- Slack unfurls that preview charts inline.
Security and governance
- Row- and column-level access through user attributes, applied across UI, AI, embeds and APIs.
- Service accounts, audit logging, usage analytics and per-user warehouse credentials.
- Custom roles with fine-grained scopes. Enterprise
- SSO, SAML and SCIM 2.0 provisioning. Enterprise
Works with the warehouse you already run
- BigQuery
- Snowflake
- Databricks
- Redshift
- PostgreSQL
- Trino
- ClickHouse
- Athena
- DuckDB / MotherDuck
Hosted cloud, self-hosted with Docker or Kubernetes, or on-prem.
Built for everyone who touches a number
Head of Data
“Stop being a dashboard factory. Define it once, let everyone else build on it.”
- Version control, PR review and CI previews mean fewer production breakages.
- Coding agents do the repetitive work while humans keep the review step.
Business leaders
“Consistent numbers across every team and every tool.”
- AI answers are auditable, with the query shown and permissions respected.
Ops and marketing
“Ask in plain language and get a governed answer.”
- Ask in Slack or on any chart.
- Explore with drag-and-drop, and get the dashboard in your inbox on a schedule.
Security and IT
“AI never gets raw warehouse access.”
- It goes through the governed layer and the user's own permissions.
- SSO, SCIM, audit logs and row-level security.
Not another “chat with your data” tool
AI on a semantic layer, not raw tables
Answers are consistent, permission-aware and inspectable.
Metrics live in your Kontala account
No separate modelling language to learn and no extra tool to keep in sync.
One layer, many surfaces
Dashboards, Explorer, agents, Slack, Teams, MCP and APIs all return the same numbers.
Warehouse-native
Queries run in your warehouse, with no extract or copy to maintain.