Data preparation startup Prophecy Inc. today launched DataGPT, a new service that allows business analysts to run complete data workflows from inside OpenAI Group PBC’s ChatGPT, Anthropic PBC’s Claude and Microsoft Corp.’s Copilot.

Each answer arrives with a visual map of the steps that produced it. The service goes after a workaround that has become common in finance and operations teams: An analyst uploads a spreadsheet to a chatbot, often without sign-off from the information technology department, and asks it to reconcile a quarter of transactions. The chatbot answers by writing a script of its own.

Most people in those jobs cannot read code, according to Prophecy, so they end up taking the numbers on trust. Closing the chat wipes the logic, along with whatever revenue definitions or approved exceptions the user explained during the session. Prophecy cites a January forecast from Gartner Inc. that half of all organizations will adopt a zero-trust posture for data governance by 2028 as unverified data generated by artificial intelligence spreads.

Raj Bains, Prophecy’s founder and chief executive, said software developers already have AI that helps them write code and take it all the way to production. “Business data users got question answering, and nothing changed,” he said. DataGPT is meant to put those users on the same footing, with the workflow behind every result in a form they can verify and change.

In a single DataGPT conversation, an analyst can load a messy Excel file and cross-check it with an audit list and records held in SharePoint or Snowflake Inc.’s data cloud. A pricing rule for discounted product bundles might come next. Prophecy’s own walkthrough ends with that workflow set to run again every month without anyone rebuilding it.

No workflow goes on a schedule until a person has looked it over. DataGPT lays out each stage on screen beside its results, and an analyst who spots a problem can fix that stage before approving the whole thing.

Every workflow is backed by open code in Git. Because reruns follow that code exactly and never go back to the AI model, they carry no added AI cost. Signed-off workflows also feed a knowledge graph Prophecy builds for each customer from its own business rules and data catalogs.

Prophecy is also pitching DataGPT as a way to bring shadow AI under IT’s control. Uploaded files never sit inside the chat. They go through Prophecy to the organization’s own data platform, where IT staff and the people who run that platform can see every workflow and govern it like anything else they manage. Scheduled runs happen on Snowflake, Databricks Inc.’s data platform or Google LLC’s BigQuery, whichever the customer already owns.

DataGPT is available today through the connector marketplaces in ChatGPT, Claude and Copilot. Companies that have not yet approved those AI tools can run it in Prophecy’s own app. Individuals can start with a free account, and the product can also be rolled out centrally by an administrator.

Prophecy has raised more than $100 million since its founding in 2017, most recently a $47 million Series B extension led by Smith Point Capital in January 2025. SAP SE, HSBC Holdings plc and Pfizer Inc. are among the customers it names.

Image: Prophecy