Rick Underwood
PhoenixAI — known until today as CelerData — has relaunched as the Agentic AI Database and named Rick Underwood as President. The Menlo Park company is staking out a position at what it argues is the next architectural inflection point in enterprise data: a category built not for human analysts, but for the AI agents now consuming enterprise data at a scale and speed analytical engines were never designed to serve.
The timing is sharp. A Cisco survey reported by VentureBeat in April found that 85 percent of enterprises are running AI agent pilots — and only five percent have moved those agents into production. Cisco framed the 80-point gap as a trust problem. Inside data infrastructure teams, another factor is increasingly being named: the database itself.
The analytical engines most enterprises rely on were architected for a different world — one of a few hundred analysts running dashboards and the steady rhythms of business intelligence. AI agents query in a different register. Thousands run at once. They expect live data, not yesterday’s snapshot. And they don’t tolerate the brittle pipelines that have been duct-taping enterprise data stacks together for years.
That is the opening PhoenixAI is built to fill.
Underwood, who joins as President, most recently served as CEO of Clumio, the cloud data protection company he took through its acquisition by Commvault (NASDAQ: CVLT).
Earlier, he was VP of Sales for Snowflake’s West Major Accounts business from 2019 through 2022, leading the region through the company’s record 2020 IPO and earning Snowflake’s Sales Leader of the Year award in FY20 — a year the company grew revenue 174 percent. Before Snowflake, he led sales at Wavefront (acquired by VMware in 2017) and Semmle (acquired by Microsoft in 2019).
That track record is, in many ways, the point. PhoenixAI is making a category bet — and category bets get won by operators who know how to scale enterprise go-to-market through inflection moments. Underwood has done it three times.
“Most of today’s analytical databases were architected for a world that no longer exists, where humans ran dashboards on flat tables and complexity was someone else’s problem,” Underwood said in a statement. “When thousands of agents need to query, reason, and act on live enterprise data simultaneously, the database is either the bottleneck or the breakthrough.”
PhoenixAI organizes its enterprise pitch around three pillars: speed, simplicity, and governance. Sub-second queries on live data with the concurrency agents demand. A unified engine for real-time and historical data — no streaming layer to babysit, no nightly batch jobs to maintain. SOC 2-grade controls, role-based access, encryption, and deployment in the customer’s own cloud (BYOC) or self-hosted via PhoenixAI Anywhere — the kind of sovereignty profile that has been stalling agentic deployments inside regulated industries.
The early customer roster — Pinterest, Expedia, and Demandbase — skews toward enterprises with both deep analytical demands and active agent strategies. That is, in other words, the precise wedge PhoenixAI intends to expand from.
The deeper question is whether the broader enterprise market will converge on the thesis that the database itself has to be rebuilt for agents, rather than retrofitted for them. PhoenixAI is betting it will. And it is betting that the team it is putting in place — anchored by a President with three prior enterprise exits on his résumé — is the one to lead the conversion.
PhoenixAI is available today. Learn more at phoenixdata.ai