Databricks’ Genie AI Agent Gives CEO Ali Ghodsi Unvarnished Truth

Databricks on Wednesday unveiled an updated version of its AI agent Genie, expanding it from a chat-based assistant into an autonomous tool capable of completing multi-step business tasks with limited supervision. The upgrade relies on an ontology built by Databricks to connect Genie with the data, workflows and business context stored on its platform, allowing the agent to reason across enterprise information and carry out longer-running jobs. Databricks Chief Executive Ali Ghodsi demonstrated the agent by walking through his own use of Genie internally. He showed how the agent analyzed employee satisfaction survey results where Databricks performed poorly, evaluated potential fixes proposed by executives and flagged weak or unnecessary options. Ghodsi also used Genie to summarize a difficult sales pitch with a prospect, including the specific product features the prospect said it needed before moving forward. Before a podcast interview, he had the agent review his prior appearances and generate likely questions. During the interview itself, Ghodsi used Genie to suggest how he should respond. The company presented Ghodsi’s usage as an example of the agent’s value for leaders at large organizations. Ghodsi referenced Dunbar’s number, the social theory that humans can maintain stable relationships with roughly 150 people, to argue that a CEO of a company with 12,000 employees cannot rely on informal channels to hear uncomfortable problems directly. Genie, he said, can surface negative signals, failed deals and internal friction that employees may be less willing to escalate in a larger company. Databricks is pitching Genie as a business-side counterpart to the coding agents that have gained traction among developers. To address concerns that autonomous agents could increase enterprise AI spending, Ghodsi said customers can manage costs through Databricks’ Unity Gateway by setting usage limits and substituting less expensive open-weight models when appropriate. The launch underscores Databricks’ effort to position itself not only as a data and AI platform, but as a provider of operational agents that can work directly across enterprise decision-making.