A New Home for the Layer
CameoDB has a second address. CameoDB.ai is where the engine meets the business: implementation consulting, real use cases, and a team that puts your data behind the decisions you care about. This site, cameodb.com, stays the home of the engine itself, with downloads, documentation and release notes.
The idea behind both is the same line you see at the top of every page: the intelligence layer between your data and your AI.
AI Doesn't Fail for Lack of Intelligence
It fails for lack of context. Three things get in the way, and anyone who has tried to put a model in front of real company data has met all of them:
- Scattered. What a business knows lives in dozens of systems. No model can reason over data it can't reach.
- Guessed. When context is missing, models fill the gap with something plausible. Plausible is not good enough for a board deck or a budget.
- Unaccountable. An answer without a source can't be checked, defended or audited, so it never makes it into a real decision.
CameoDB answers all three with one layer that sits between your sources and every AI your business runs.
One Layer, Three Jobs
Index Everything
Text, numbers, dates and identifiers are all fully indexed. An agent can filter by date or amount, search free text, and look a record up by its id, which takes a direct key-value path. That makes pivoting from one record to the next instant.
Describe What It Means
Every index and field carries its type, its capabilities and a plain-language description. describe_index hands all of it to the agent, so your AI learns what the data is, not just what the columns are called, and builds the right query the first time.
Serve Grounded Answers
Agents query through MCP, across many indexes in a single call with search_across_indexes. Every hit is tagged with the index it came from, and a field projection returns only the fields the question needs, which leaves the context window room to reason. The engine behind it is the same single binary you can download today and connect to Claude Code, Claude Desktop, Cursor or Windsurf in a minute, as the Build page shows.
Built In, Not Bolted On
Data engineering lives inside the database. Point CameoDB at a CSV, TSV or JSON export, local or over HTTPS, and it detects the structure, picks the types and loads it, with no separate pipeline to build. In practice that means a first working session ends with your own data already running as a knowledge base your AI can query, and your team adds the next source without help.
Built So Every Answer Can Be Defended
- Grounded, not guessed. Answers come only from what the layer returns. Nothing gets filled in from a model's memory.
- Every answer has a source. Each hit names the index it came from, so every claim can be cited.
- Read-only by design. Every MCP tool is a read. The AI retrieves and synthesizes, while loading data runs separately, under your control.
- Partial is never whole. If a source can't be read, the answer says so instead of passing off half the picture as the full one.
These are not policies layered on top. They are how the engine answers: every MCP session opens with the same instruction to answer only from what the tools return, and every response says exactly what it covered.
Where It Makes a Difference
The use cases on CameoDB.ai start from the questions people actually ask:
- Executive decision support. Briefings assembled from live company data, with every claim traceable to its record. Which accounts are most at risk this quarter, and why?
- Planning and forecasting. Ground plans in what actually happened across sales, finance and operations. What drove the variance between plan and actuals in Q3?
- Threat intelligence. Pivot across indicators, hashes and reports in one federated query, with provenance on every hit. Which files share this signer certificate, and which domains did they reach?
- Incident and operations. Correlate tickets, logs and changes so responders start from evidence. What changed before the outage began?
- Compliance and audit. Read-only retrieval with cited sources that auditors can verify line by line. Show every contract with a non-standard liability clause.
- Customer intelligence. One view of each customer across CRM, support and usage. Summarize everything we know about this account.
Each one grows the same way: add an index on the same server as a new source arrives, and add servers when the data outgrows one. A single distributed query still returns one merged, cited result.
We Don't Sell You a Database
We install your intelligence layer, then hand you the tools. A CameoDB.ai engagement runs in four phases:
- Discover. Map your sources and the decisions you want AI to support, and agree on the use cases that matter first.
- Implement. Ingestion, schemas, field descriptions and MCP wiring to the AI tools your teams already use.
- Enable. Use cases go live with your people, who learn to ask better questions and to trust the answers.
- Evolve. New sources, new agents, new decisions. The layer grows with the business.
Prove It Once, Then Own It
The model behind it is simple. We prove it in a first working session that turns your own data into a business knowledge base. You own it: the platform and tooling to add every new source yourselves. We support it with an enterprise subscription for support, upgrades and maintenance. And we extend it with consulting on request, for the specific and complex use cases.
On the product side that lines up with the channels on the downloads page: the same engine everyone can run, an Enterprise channel with a support agreement and a maintenance cycle, and a Managed channel that runs CameoDB for you, in your own data centre or in the cloud.
Give Your AI Something Worth Knowing
Tell us which decisions you want your data behind, and we'll show you the layer that gets it there. The first step is a working session on your sources and use cases.
Visit cameodb.ai, or write to cameodb.ai@gmail.com. If you'd rather start with the engine itself, the Quickstart has a node running and a dataset loaded in a few commands.