The platform
Everything a data team does, on one DuckDB workspace.
ZAANIX runs a DuckDB engine for each workspace inside its own server. Around it: the tools to connect data, query and model it, build what people use, and govern who sees what — for people, AI and agents alike.
Query and explore
Files, tables and remote objects, queried where they are, with everything you need around the SQL.
Getting started
SQL workbenchTabs that persist, a file and table explorer, saved queries in folders, and the history of every query you ran.
Results, five waysThe grid, a chart, a pivot table, a column profile and the measured query plan — for the same result.
Errors you can act onThe failing line highlighted, the message in words, and "Fix with AI" when you want it.
Table pagesOverview and profile, columns, checks, lineage and access for any table or file; links into lineage.
Interactive explorationCross-filtered views of any result: brush one chart and the rest follow (Mosaic).
Compare two datasetsRows added, removed and changed between two tables or files, column by column.
Workspace searchOne box for tables, columns, files, queries, dashboards, notebooks, metrics and apps.
Files from anywhereA location browser for folders and buckets, or your system's own file dialog when ZAANIX runs on your machine.
DashboardsGrid dashboards of KPIs, charts, tables and maps, or cross-filtered Mosaic dashboards with inputs.
SQL notebooksSQL, text and input cells; each cell can read the one above by name, and outputs are saved.
Data appsStreamlit, Dash and Gradio apps on workspace data, run as processes, in Docker or on Kubernetes — or in the viewer's browser.
Publishing with reviewApps are published after an administrator's review, on their own origin, scaled to zero when idle.
Alerts and snapshotsSQL alerts with triggered and resolved states; dashboards and apps rendered to PNG or PDF on a schedule.
Delivery channelsSlack, Microsoft Teams, email, PagerDuty and signed webhooks, behind an egress guard.
Signed embedsDashboards and notebooks inside your own application, with rows per tenant through access policies.
Queries as HTTP APIsPublish a read-only query as an endpoint with typed parameters, its own key and a rate limit; JSON or CSV back.
TemplatesStart a workspace from a ready-made set of tables, queries and dashboards.
Comments and mentionsThreads on notebook cells, dashboards and tables, with an inbox and email.
Connect
35 kinds of source, connected once and used by every workspace. Credentials are encrypted and write-only.
Connections and syncs
Object storageAmazon S3, Cloudflare R2, Google Cloud Storage and Azure Blob — Parquet, CSV, JSON, Delta and Iceberg, queried in place.
Lakehouse catalogsAWS Glue, S3 Tables, any Iceberg REST catalog (Polaris, Nessie, Lakekeeper) and Databricks Unity Catalog.
DatabasesPostgreSQL, MySQL, SQLite and other DuckDB files, attached read-only.
WarehousesSnowflake, BigQuery, Redshift, ClickHouse and Microsoft Fabric.
SaaS applicationsSalesforce, HubSpot, Stripe, GA4, Airtable, Notion, GitHub, Jira, Zendesk, Shopify, Intercom, Linear, Pipedrive and Mailchimp.
Google Drive and SheetsSign in with a Google account, or use a public Sheets link; HTTP and REST endpoints too.
Scheduled syncsFull, incremental or append loads on a schedule, with transformations checked before they save.
Streams and change data captureKafka topics, Kinesis streams and HTTP pushes appended as events arrive; Postgres tables mirrored through logical replication, or Debezium events.
Reverse ETLQuery results sent to databases, files or APIs — only what changed — and written back to Iceberg and Delta tables.
Workspaces anywhereA workspace is a DuckDB file in any folder on the server, or an object in S3, R2, GCS or Azure kept in sync.
Model and trust
Definitions people agree on, and checks that say when the data is wrong.
Models, metrics and qualitydbt projectsReal dbt Core compiles against a shadow of the workspace; ZAANIX runs the SQL in the workspace's engine. Save any query as a model.
Semantic layerMetrics defined once (MetricFlow-shaped, imported from dbt) and computed the same way in the explorer, by the AI and by agents.
Questions from metricsAsk "revenue by region last quarter" and the answer is computed through the semantic layer, not guessed.
Data qualityChecks suggested from the data — nulls, duplicates, categories, ranges, joins, freshness — on a schedule, with the failing rows one click away.
Metric monitorsUnusual days, weeks and months flagged, with the segments that drove the change, delivered to a channel.
Schema and freshness watchesKnow when a table's columns change or its data stops arriving.
Prepare dataStep-by-step recipes — filter, rename, cast, fill, clean text, split, deduplicate — previewed step by step and saved as a view, a table or a dbt model.
Find joinsHow tables join — declared keys and relationships inferred from names — confirmed on the data: coverage and the rows an inner join would drop.
Row filters and column masksPer table, per role or team, enforced by rewriting every query with DuckDB's own parser — for people, BI tools, the AI and agents.
Catalog and lineageDescriptions and tags that people and AI read; a graph from sources to consumers; OpenLineage events.
Personal dataFind columns that hold personal data, tag them, and mask them with a policy.
Sharing and teamsWorkspaces shared with people or teams as owner, editor or viewer.
Single sign-on and SCIMOIDC with identity-provider groups as teams; SCIM creates, updates and deactivates people.
Audit exportThe audit log streamed to Splunk, Datadog, Elasticsearch, a webhook or a bucket, in order.
Version historyEvery save of notebooks, dashboards, queries, metrics and dbt projects, with diffs and restore.
Git syncPush the workspace's work to a Git repository, review it there, and pull changes back.
An AI panel on every pageIt sees the table, result, dashboard or error on screen; its answers become SQL tabs, charts and widgets you can edit.
Builds dashboards and appsA checked build plan from a question, created with one click.
Your modelAnthropic, OpenAI, Amazon Bedrock and Bedrock agents, AgentCore, Ollama — set for the server, or brought by each person.
ZAANIX AgentA separate app for missions: context, a plan, queries as you, approvals and briefs. How it works.
MCP server94 tools for Claude, Cursor and other MCP clients over stdio, SSE or Streamable HTTP; a REST API with OpenAPI for the rest.
ApprovalsA statement from an agent that changes data waits for a person, in ZAANIX or in the agent's own client.
Agents ZAANIX runsReady-made agents from a marketplace, run on a schedule with read-only tools.
Agent2AgentOther agents call ZAANIX's agents over A2A, and ZAANIX asks remote agents.
Workspace managementAn administration list with bulk actions, a create wizard with clone and templates, and a detail page per workspace.
Backups and bundlesBackups and restore, with a safety backup first;
.zaanix bundles to move a workspace.Quotas and policiesStorage and engine quotas, an idle policy, and rules for who may create workspaces.
Cluster modeSeveral nodes behind one load balancer, with PostgreSQL metadata.
The Postgres protocolTableau, Power BI, Metabase, psql and Postgres drivers query workspaces directly, with the same policies.
OrchestrationAirflow operators, Dagster resources and Prefect tasks start ZAANIX work and wait for it.
Result cacheRepeated profiles and queries answered from a cache that knows when the data changed.
Usage and costCompute time, AI tokens and storage by person and workspace, priced at your rates, with budgets that notify a channel.
ObservabilityPrometheus metrics, OpenTelemetry traces and structured logs.
ThemesDark and light themes, fonts and a UI scale, per person.
See it on your own data.
One container; the first person to open it creates the admin account. Talk to us and we will set it up with you.
Available on request
Bring a sample of your own data: a folder of files, a database or a warehouse. We will get ZAANIX running on it and show you around.
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.