For Snowflake
Data Workers runs the detect, diagnose, fix, review and verify loop on a Snowflake estate when nobody is logged in. A task that stopped days ago is found and diagnosed, a fix is proposed, a named human approves anything irreversible, and every applied change carries a receipt with its blast radius across downstream models and dashboards.
For a data team already running Snowflake: warehouses, tasks, streams and dbt, with an on-call rota and more incidents than people.
Last updated: September 10, 2026 · Dhanush Shetty, founder, Data Workers
None of these is a Snowflake defect. They are the failures a busy estate produces, and the hours between one of them starting and somebody verifying a fix are what this platform is for.
A task that stopped running days ago and nobody noticed
A task in the graph suspended after repeated failures. Everything downstream kept querying the table it should have been refreshing. The first person to notice was whoever read the number and knew it was wrong.
A warehouse burning credits for a consumer who left
A scheduled query has run hourly for eleven months. The dashboard it fed belonged to someone who changed teams. Nothing is broken, which is exactly why nobody has looked at it.
A stream that went stale past its retention
The consuming task fell behind, the stream's offset aged past the source table's data retention, and the stream became unreadable. Recovering it is a reload, and knowing that early is the difference between an hour and a day.
A dbt model that compiles clean and ships wrong numbers
The SQL is valid, the tests pass, and a join fans out because an upstream table quietly stopped being unique on its key. Nothing fails. The report is wrong by a factor nobody can explain in the meeting.
Alerts are not fixes. Detection tells you the first of those sentences. The rest is the work.
Runs the loop
Autonomous Data-Conductor
The Conductor watches task outcomes, stream lag and freshness and starts work without waiting for somebody to log in. On a suspended task it reads the failure history, the objects the task touches and the lineage around them, forms a diagnosis, and produces the change it believes fixes it. An autonomy dial governs how far it goes on its own: at the lowest setting it stops at a proposal, higher up it may run the dry run and stage the change, and anything irreversible always waits for a named approver regardless of the setting. The dial is a policy you set, not a claim about what it has already done unattended in production.
Does the work and writes
Data-Agents Swarm, 20 specialised agents
On Snowflake the incident debugging, data quality, schema evolution and data change review agents carry most of the load. The cost and cleanup agent has an obvious job here, finding warehouses and scheduled work running for consumers who are gone, and its output is a list of specific objects with the evidence for each, not a savings figure we would have no basis to print.
What the agents reason on
Data Context Wizard
The Context Wizard holds ownership that is actually acted on, metric definitions, PII tags, column-level lineage and the record of past corrections, every fact provenance-stamped so an agent can tell what it learned from where. Cross-cloud matters even to a team that considers itself all-in on Snowflake, because the wrong numbers usually arrive from the edge of the estate: a vendor export, a Postgres, a BigQuery dataset that somebody landed once and nobody owns.
Where humans approve and audit
Spellbook Data Catalog, in preview
Spellbook is the catalog the agents write and keep current, and the plane where a human approves a change and reads back what happened. It sits alongside Snowflake Horizon rather than replacing it: Horizon stays the governance surface. Spellbook is in preview, so treat it as a direction rather than a shipped surface.
What each module has to do to qualify as this kind of product, stated generically so you can use it on any vendor: the autonomous agentic data platform.
The first three steps need no form, no account and no key we issue. The open-source core is Apache-2.0: 11 agents and 160+ MCP tools that read, analyse and recommend.
The full list of who we are not for, including the cases where the honest answer is to buy something else: should you use Data Workers.
Does this replace Snowflake Horizon?
No. Horizon remains the governance surface on Snowflake. The Data Context Wizard reads what is there and adds what a governance catalog does not hold: metric definitions, acted-on ownership, past corrections and cross-cloud column-level lineage, provenance-stamped. Spellbook, which is in preview, is the human approval and audit plane for agent-made changes.
Will it reduce our Snowflake bill?
The cost and cleanup agent finds warehouses, tasks and scheduled queries that run for consumers who are gone, and reports each one with the evidence. Whether that reduces your bill depends on what it finds in your account, which we cannot know in advance. We are pre-revenue and publish no savings figures, because we have no customer runs to derive one from.
Can it change production on Snowflake without us?
Not for anything irreversible. The write path is propose first, dry run where one exists, and a named human approves before an irreversible change lands. Every applied change carries a receipt: the diff, the approver, the blast radius across downstream models and dashboards, and the rollback path.
Two next steps
Run the read-only agents on Snowflake yourself, free: clone the open-source core.
Or bring one real incident to a 45-minute session and watch the approval gate stop the agent: book a time. Pilot Program $7,500 one-time, then Scale from $1,000 per month, no usage meter, as published on 10 September 2026. Pricing.