Contextbase keeps semantics, knowledge, memory, traces, and evaluations in one governed system that every agent can read from and write to.
MENLO PARK, Calif., Nov. 5, 2026 — PhoenixAI has launched Contextbase, the context infrastructure for AI agents. The company unveiled it at NEXT QUERY 2026, its Data + AI Infrastructure Summit. Contextbase stores semantics, knowledge, memory, traces, and evaluations as context in one system, so agents understand, reason, and act better with every run.
An agent only knows what is in its context. What the data means, what the business knows, what earlier runs learned, what this run did and whether it worked all reach the agent as context, or not at all. In most stacks today those pieces live in separate tools: definitions in a semantic layer, documents in a vector store, memory in a memory service, and traces and evaluations in observability tools. Each keeps its own copy and its own permissions.
Contextbase holds all of it in one place. Semantic views of entities, dimensions and metrics, defined by the business in Semantic SQL. Typed, versioned knowledge records such as a spec, an incident review or a playbook, linked to each other. Memory that agents write from earlier work. Traces of every model call and tool call in a run, with prompts, outputs, tokens and cost. And evaluations, from test datasets and eval runs to scores on live traffic.
“An agent is only as good as its context, and today it has to glue that context together from separate systems on every turn. What it gets is partial, inconsistent and often stale,” said James Li, co-founder and CEO of PhoenixAI. “Context is the next great infrastructure layer, and it belongs in one real-time system, where it compounds. Every run makes it better, and every agent you ship makes the next one smarter. That’s the flywheel Contextbase starts.”
Contextbase is built into the PhoenixAI database, so context lives on the same cluster as the business data and one query can reach all of it under one set of permissions. Everything in it is linked, versioned, governed and directly readable by agents. The agent is the non-deterministic part of the system, so the place it reads from is built to be steady: definitions are verified, the system enforces policies, and records are kept as written.
PhoenixAI is building applications on Contextbase, starting with ObserveStack, for tracing and evaluating agents, so teams can see what each run did and score whether it worked, using the same store their agents read from.
Contextbase will be available in private preview on November 4th. Teams can request a free trial today at phoenixdata.ai.
PhoenixAI is the Agentic AI Database, purpose-built to give autonomous AI agents sub-second access to live enterprise data at massive scale. The platform unifies real-time and at-rest data in a single AI-native engine, combining the speed and concurrency that agentic workloads demand with the governance and deployment flexibility enterprises require to put AI into production. PhoenixAI runs in the customer’s own cloud account, or self-managed on the customer’s own private cloud, or on their own hardware, including on-premises. Formerly CelerData, and the creators of the StarRocks open source project, PhoenixAI is headquartered in Menlo Park, California.
The Big Data Inside Amazon’s New Fire Phone
The new Fire phone that Amazon launched this week looks like your ordinary black smartphone,…
Three Reasons to be Scared of the Internet of Things
We know the Internet of Things forecasts: 50 billion connected devices by 2020. Apparently, there’s…
Wanted: Intelligent Middleware That Simplifies Big Data Analytics
We’ve seen tremendous technological innovation in the data analytics space over the past 10 years….
GPUs Tackle Massive Data of the Hive Mind
LIVE from GTC12 — The flock of birds that weaves seamlessly through the sky, propelled…
Why Hadoop on IBM Power
In the quest to achieve data-driven insight, Hadoop running on Intel X86-based processors has emerged…
Datanami’s Leverage Big Data Summit Wraps Up
Dialog and networking were on the Datanami agenda this week as we kicked off our…
CrowdStrike and Dynatrace Want Security Teams to Search Data Where It Already Lives
CrowdStrike and Dynatrace have announced three new integrations aimed at helping companies cut the cost…
Who Controls Enterprise AI Agents? Data Platform Vendors Race to Provide the Answer
AI agents are everywhere. Every other news story is related somehow to AI agents. As…
The Token Trap: Why AI Costs Spiral at Scale
Most enterprise AI teams discover the token cost problem somewhere between the demo and production,…
Hugging Face Launches ProtoStar to Tackle Science’s Data Problem
There’s a lot of scientific data out there – and universities, government agencies, and research…
Dataiku Headless Tackles a Growing Problem With AI Coding Agents
Getting an AI coding assistant to write a data pipeline is becoming easier. No surprises…
Oracle’s New Lakehouse Tools Target a Blind Spot in Enterprise AI
Giving AI agents access to company data is one thing. Making sure they understand what…
