Prophecy launches DataGPT to turn AI chats into reviewable data workflows

Prophecy wants business analysts to trust more than an AI-generated answer: it wants them to approve the process behind it. The company launched DataGPT on October 28, 2026, with immediate availability in ChatGPT, Claude, Microsoft Copilot and its own app. Its promise is a visual workflow users can inspect before turning it into repeatable code.

The launch targets finance, operations and other business analysts who work with spreadsheets and desktop tools. Prophecy’s pitch is that answering a question is only part of their job. Before a report can be delivered, someone must bring in files, reconcile conflicting systems, match records and apply business rules. DataGPT is designed to handle that preparation as well as analysis and finished outputs, within a conversation.

A review step between the request and the report

An analyst describes the desired outcome, and DataGPT builds the workflow behind it. According to Prophecy, each result includes a visual representation of the steps used to produce it. Analysts can review individual steps and their results, make refinements and approve the final version. The proposed division of labor is explicit: AI builds the workflow, while the analyst checks and changes it rather than starting from scratch.

Prophecy illustrates the process with an analyst uploading a messy Excel file, checking it against an audit list and information in SharePoint, Snowflake and Databricks, then applying a pricing rule for discounted bundles. The analyst reviews the output and schedules the work to run monthly. That is a company-described example of the intended workflow, not a measured customer outcome.

The connections extend beyond uploaded spreadsheets. DataGPT supports files, SharePoint, data warehouses, enterprise resource planning systems and customer relationship management systems. Prophecy also says it uses a knowledge graph—a structure connecting business context with workflows—to help teams apply relevant definitions and rules. That feature is intended to carry business meaning into the work, alongside the instructions given in a particular conversation.

Keep the code, not just the answer

After approval, the workflow becomes code that runs on the organization’s existing data platform. Prophecy says every workflow is backed by open, editable code in Git, a system for tracking code changes. The visual review is therefore not the only way to examine the logic: the underlying code remains available for inspection and editing, rather than leaving the finished answer as the sole record of the work.

Prophecy says approved workflows can run on a schedule without generating new AI code each time. It describes those reruns as carrying no added AI cost. That claim concerns reusing the approved workflow without fresh generation; it is not a promise that the underlying data platform or the service as a whole costs nothing.

Files go through Prophecy

The company also makes a specific claim about uploaded files: they pass through Prophecy to the organization’s own data platform, rather than into an AI chat. Prophecy says IT and data-platform leaders can see the workflows and govern them on infrastructure they already use. Organizations can start with a free account, and administrators have centralized deployment options.