Everything a modern application stores,
in one place.
Data Oil is one SQL surface over everything a modern application stores: rows, documents, relationships, full-text search, vectors and translations, on data that is enriched as it arrives, queried in SQL or in plain language, against the AI model you choose. The same surface serves a person at a console and an agent in a pipeline.
Ingestion to outcome
Five kinds of data, one engine, one statement
Data Oil uses Artificial Intelligence, Machine Learning and Natural Language Processing to turn raw, unstructured data into clear, searchable insight you can interact with conversationally — an established model applied from the point of ingestion to the point of outcome.
What you ingest is analysed on the way in: the people, places and things it mentions, and how they relate, go into the knowledge graph beside it. That derived data is produced for everything you ingest unless you switch Derived data off for an upload — your choice, recorded with the job.
Send one file, or a hundred at once — a folder's worth, or a zip of it. They arrive as one submission: checked before anything starts, published together when all of it is in, and picked up where it stopped if anything interrupts it.
For non-technical people: the Data Oil console
Ask a question in plain language and receive a detailed answer in plain language, with the evidence it came from. No query language, no code.
For technical people: the API
The same data through one HTTP surface: SQL over rows, vectors, keywords and relationships in a single statement, from any language that can make an HTTP call, any PostgreSQL client, a client package, or an MCP server your agent already knows how to call.
The whole product in one statement
Filter on a column, rank by meaning, score a keyword, in one query
A write commits once and is searchable by keyword, by meaning and by relationship the moment it commits, because every index is built by the same engine over the same records. No second query language, and no job keeping a database and a search engine in step.
See the statement itselfFor the technically minded
SELECT title,
VECTOR_DISTANCE(embedding, EMBED('casing failure'), 'cosine') AS distance,
$score AS keywordScore
FROM Report
WHERE SEARCH_INDEX('Report[body]', 'casing failure') = true
AND filedAt > date('2026-01-01')
ORDER BY distance LIMIT 10;
What you get
Twelve things that are one system rather than twelve
One query
Filter on a column, rank by meaning, score a keyword and walk a relationship - all in the same statement, over the same records.
Every model of data
Rows, documents, graphs, vectors and full text, with the joins, windows and traversals you already expect from a database.
Enriched on the way in
Files, media and thirty-one document formats are read, transcribed, embedded and translated as they arrive, and stored beside the original.
Answers with the evidence
A question in plain language comes back answered, with its sources, its citations and a confidence score.
Your model, your key
You choose the AI model and register your own key. It is sealed, it never leaves, and you can see exactly what would be sent before it is sent.
Secured in layers
Rights down to a single column, the same fence on every door, sign-on with your own directory, and an audit line for every act.
Reliable by construction
Every durable byte in object storage, nodes replaceable, and backups proved by real restore drills on the cadence you choose.
Twelve regulatory regimes
IRAP, CSA STAR, FedRAMP, GDPR, HIPAA, ISO/IEC 27001, PCI DSS, MTCS, ENS, SOC 1, 2 and 3 - declared per job and per column, and enforced.
Your cloud or ours
Oracle Cloud, AWS, Azure, Google Cloud or your own storage - several at once, with residency enforced on every write.
Every door you already use
HTTP and JSON, the PostgreSQL wire protocol for your BI tools, packages for .NET, Python and Node, and an MCP server for your agent.
Described, for people and for agents
Every database, collection and column carries a description, so a question is answered about what the data means, not only what it is called.
A console for your administrator
Roles, sign-on, backups, restores and spending - your own administrator runs your own tenant, and every act they take is in the audit.
All of it on one published unit rate, on the pricing page.
Eight things that are true of the build
What the architecture buys you
- Modular Ingestion & Refinement
- Persona-Aware Insight Delivery
- Real-Time, Scalable Architecture
- Compliance-Ready Commentary Engine
- Plug-and-Play Modularity
- Edge computing in your own system
- Encrypted end to end
- No code required
The long form of every one of these is in Data Oil in detail — see the manuals for further details.
Next
One published number, and a bill you can check.
Two cents per thousand units, 115 meters behind it, every rate visible before and after it is charged — and what the work actually costs checked against your budget at every stage, stopping above 95% of it.
The pricing Security & compliance