ZAANIX

Your data workspace, with an agent that works in it.

ZAANIX brings your files, databases, warehouses and lakehouses into one self-hosted workspace on DuckDB. Query, model, chart and govern the data yourself — or hand the question to ZAANIX Agent, which works with your access and asks before it changes anything.

Talk to us How it runs

New in 3.0.0: five sections, inline AI, scheduled missions, and a platform that works with no model

You drive. SQL, dashboards, notebooks and apps on the same data — a cross-filtered sales dashboard here.

A Mosaic dashboard in ZAANIX: monthly sales by category, orders by channel, revenue by region and the order table, with filters above
35kinds of source: buckets, lakehouses, databases, warehouses and SaaS apps
94tools, the same for the agent, MCP clients and the REST API
1DuckDB engine per workspace, in the server's own process
AES-256encryption for every stored credential; secrets never reach the browser
Two ways to work, one workspace

The platform for doing it yourself. The agent for handing it over.

Both work on the same workspaces, the same data and the same permissions. Start in either; anything the agent makes opens in the platform.

ZAANIX

For analysts and data engineers who want the tools in their hands.

  • A SQL workbench with results, charts, pivots, profiles and measured query plans
  • Dashboards, notebooks and data apps — Streamlit, Dash and Gradio — on live data
  • Models, metrics and quality checks: dbt projects, a semantic layer, tests on a schedule
  • Governance: row filters, column masks, lineage, audit and SCIM
Everything in the platform

ZAANIX Agent

For anyone with a question, or work they would rather hand over.

  • Missions: say what you need; the agent plans, queries and shows its work
  • Context you choose: tables, server files, cloud objects, lakehouse tables, uploads
  • Approval where it matters: reads run; changes wait for you, by your autonomy setting
  • Results you can keep: pin to a dashboard, save the query, turn a mission into a brief
How the agent works
What is in it

Everything a data team does, in one self-hosted workspace.

Query and explore

Files, tables and remote objects, queried where they are.

  • SQL workbench with tabs, history, formatting and a measured plan
  • Data explorer: profiles, column detail, pivots, compare two datasets
  • Interactive exploration: cross-filtered views of any result
  • Workspace search across tables, columns, queries and dashboards

Build and share

From a query to something other people use.

  • Dashboards: grid or cross-filtered, with maps, alerts and snapshots
  • SQL notebooks where each cell builds on the one above
  • Data apps: Streamlit, Dash and Gradio, served next to the engine
  • Signed embeds and queries published as HTTP APIs

Connect

Once, for every workspace.

  • Storage and lakehouses: S3, R2, GCS, Azure, Iceberg, Glue, Databricks
  • Databases and warehouses: Postgres, MySQL, Snowflake, BigQuery, Redshift
  • SaaS: Salesforce, HubSpot, Stripe, GA4, GitHub, Jira, Shopify and more
  • Streams and CDC: Kafka, Kinesis, HTTP pushes, Postgres replication

Model and trust

Numbers people can rely on.

  • dbt projects compiled by real dbt Core, run in the workspace's engine
  • Semantic layer: metrics defined once, used by people, AI and agents
  • Data quality: suggested checks on a schedule, failing rows one click away
  • Metric monitors that flag unusual days and say which segment moved

Govern

Access that holds for every query, whoever asks.

  • Row filters and column masks, applied by rewriting each query
  • Catalog and lineage from sources to dashboards and apps
  • Personal data: find it, tag it, mask it
  • SSO, SCIM, audit export, version history and Git sync

Automate and integrate

ZAANIX in the rest of your stack.

  • The Postgres protocol: Tableau, Power BI, Metabase and psql connect directly
  • Orchestration: Airflow, Dagster and Prefect run ZAANIX work
  • Reverse ETL into databases, APIs, Iceberg and Delta tables
  • MCP and A2A for your own agents, and a REST API with OpenAPI

See the whole platform

SQL first, with everything around it.

Every workspace is a DuckDB database. Files on the server, objects in your buckets and tables in your lakehouse are queryable where they sit.

  • Results, a chart, a pivot, a profile and the query plan in one place
  • Errors that point at the line, and a fix from the AI when you want one
  • Statements that change data wait for approval before they run
The ZAANIX SQL workbench: a monthly revenue query and its bar chart, with the data directory and tables on the left

Notebooks that explain, not just compute.

SQL, text and input cells. Each cell can read the result of the one above by name, so an analysis reads top to bottom — and keeps its outputs.

  • Inputs that parameterise every query below them
  • Comments on cells, version history and Git sync
  • Export to Markdown; agents can create and run notebooks too
A ZAANIX SQL notebook analysing sales transactions, with a text cell, an input and a SQL cell

Know where every number comes from.

Lineage is built from the SQL itself — syncs, views, dbt models, dashboards, alerts and apps — with DuckDB's own parser.

  • Trace a table to its sources and to everything built on it
  • Descriptions and tags that people, the AI and agents all read
  • OpenLineage events for the catalog you already run
ZAANIX lineage: a sales file feeding a staging view, four analysis views, a Streamlit app and a Mosaic dashboard
ZAANIX Agent

Hand over the question. Keep control of the answer.

ZAANIX Agent is a separate web app next to the platform. You sign in with your ZAANIX account; the agent works with your access, uses the platform's 94 tools, and holds every change for you.

  1. Choose the contextTables, files on the server, cloud objects, lakehouse tables, or files from your computer.
  2. AskPick what you want — analyse, build a dashboard, check quality, investigate — or just write it.
  3. It works as youA plan, then queries ZAANIX runs with your permissions. Every step can be opened.
  4. You decide what staysApprove changes, pin results to dashboards, save queries, or turn it into a brief.
The ZAANIX Agent home: a prompt, capability chips such as Analyse data and Build dashboard, sample prompts and the Context panel
Start from a capability. Sample prompts are written for the data in your context.
A ZAANIX Agent brief: a report page with the summary, key findings and a chart
Keep it as a brief. An editable report that refreshes its numbers and saves as a dashboard.

How the agent works

Built to be trusted

Your data stays where you put it.

Self-hostedOne container, on a laptop, a server or Kubernetes. Nothing leaves unless you connect it.
One set of permissionsPeople, the AI, agents, BI tools and APIs all go through the same roles, row filters and column masks.
Changes wait for a personAn agent's statement that changes data, publishes or reaches outside is held until someone approves it.
Everything is on the recordAn audit log you can stream to Splunk, Datadog or a bucket; version history for every notebook, query and dashboard.
Credentials never reach the browserConnection secrets are encrypted at rest (AES-256-GCM) and write-only once saved.
Your model, your choiceAnthropic, OpenAI, Bedrock, OpenRouter, Ollama or any OpenAI-compatible endpoint — or none at all.
Works with

Fits the stack you already have.

BI tools
Tableau, Power BI, Metabase, psql and any Postgres driver, over the Postgres protocol
Transformation
dbt Core projects, imported MetricFlow metrics, dbt tests next to ZAANIX's checks
Orchestration
Airflow operators, Dagster resources and Prefect tasks
Agents
Claude, Cursor and other MCP clients; Agent2Agent (A2A) in both directions
Delivery
Slack, Microsoft Teams, email, PagerDuty and signed webhooks
Identity
OIDC single sign-on, IdP groups as teams, SCIM provisioning
Code
Notebooks, queries, dashboards, metrics and dbt projects synced to a Git repository

Running where your data is.

One container on a laptop, Docker Compose for a team, Kubernetes or a cluster for more. Nothing else to install, and no network needed at runtime.

Deployment options Read the docs
  • On a laptop. One container; the first person to open it creates the admin account.
  • For a team. Docker Compose with PostgreSQL metadata, ZAANIX Agent, a local model and Prometheus as you need them.
  • At scale. Kubernetes manifests, and a cluster mode with several nodes behind one load balancer.

Talk to us about ZAANIX.

ZAANIX is available on request. Tell us about your data and your team, and we will set you up with access, the deployment files and a walkthrough.

Email us Message on LinkedIn

or write to ajmalnazirbaba@gmail.com