PhoenixAI (formerly CelerData) (@phoenixdataai) / X

PhoenixAI (formerly CelerData)

1,065 posts

PhoenixAI (formerly CelerData)

@phoenixdataai

The Agentic AI Database. Giving autonomous AI agents sub-second access to live enterprise data at massive scale. Built for production. Formerly CelerData.

Menlo Park, CA

Joined August 2022

PhoenixAI (formerly CelerData)
@phoenixdataai
Aug 7
People tolerate latency. Agents don't. That one difference is quietly reshaping what "real-time" has to mean. Our new post breaks down the full loop (ingest, update, query) and what comes next: AgentBase. Agents are the new power users. Made for them → bit.ly/4vzTBhO
PhoenixAI (formerly CelerData)
@phoenixdataai
Aug 6
After you write a row, how long until queries see it? Our worst case used to be ~20 seconds. Now: sub-second p99 at ~1M rows/sec — and <5s worst-case in a production accounting workload restating years of history. The full path: bit.ly/4ypxQD0
PhoenixAI (formerly CelerData)
@phoenixdataai
Aug 5
When the board asks how your AI agents are governed, "give us a few weeks" is the wrong answer. PhoenixAI keeps the evidence ready: audit logs, row-level security, masking, SOC 2, BYOC. Agents: sub-second answers. Governance: a paper trail. bit.ly/4gAYgvd #ai #data
PhoenixAI (formerly CelerData)
@phoenixdataai
Aug 4
Your database's busiest user never sleeps. AI agents query 24/7 at machine speed — the data layer either keeps up or becomes the bottleneck. PhoenixAI: sub-second on 100s of billions of rows, 10K+ QPS. Coinbase runs 573B rows in production. bit.ly/4gAYgvd #ai #data
PhoenixAI (formerly CelerData)
@phoenixdataai
Aug 4
Missed our webinar? The recording is up 👉 🎥 bit.ly/450EwKr See what updates actually cost on columnar storage, why copy-on-write and merge-on-read both fall short, and the architecture that keeps data queryable within 5 seconds — while query performance stays flat.
00:00