The one thing that fixes
your AI data problem.
For C-Suite leaders who need certainty, not complexity.
The problem
The problem you already feel — even if you can't name it
Your organization is using AI. Or trying to. And somewhere in the process, someone raised a hand and said: “Wait — where does our data actually go when we ask it a question?”
Nobody had a clean answer.
That is the problem Data Oil solves. Not a technology problem. A control problem. A cost problem. A governance problem. And right now, probably a sleep problem.
Where you are
Four things that are true about your current setup
-
You are paying three bills to do one job.
Most organisations running AI on their own data are maintaining a database, a search system, and an AI memory layer simultaneously — keeping them synchronised, paying three vendors, and employing engineers whose entire job is making sure the three agree with each other. Data Oil is one system that does all three. One bill. One relationship. One source of truth.
-
You don't actually know what leaves the building.
When someone in your organisation asks your AI assistant a question, data leaves your environment and travels to an AI model provider. Right now, can you tell your board — precisely, documented — what was sent? Data Oil shows you exactly what would be sent to your AI provider before it is sent. Your security team will ask this question. Data Oil means you already have the answer.
-
Your pricing is unpredictable.
Data costs at scale are notoriously difficult to forecast. Bulk processing jobs surprise finance. Bills don't reconcile to the work that caused them. Data Oil prices every operation on one published rate card, and checks what each job has actually spent against the budget you set at every stage — work stops above 95% of it, without anyone having to watch. Finance can reconcile every charge to a dated record, to the team that incurred it, without a spreadsheet.
-
“In production” is a claim most vendors can't make.
Data Oil is not a roadmap promise or a beta feature. Three businesses are running it in production today — one of them uses it for all of its reporting. When your team asks whether this has been proven, the answer is yes, with names.
Governance
The six things you can say to your board — with the proof behind each
Governance conversations with boards tend to circle the same territory. Here is how Data Oil answers each:
Our AI uses our model and our key — not the vendor's.
What you can saySealed, and assigned per database
Data Oil holds no AI model of its own. Your organisation's AI endpoint and your own API key are sealed and assigned per database. You are not sharing model access with other organisations.
We know what leaves our environment, documented.
What you can sayBefore it is sent, not after
Every query shows exactly what would be sent to the AI provider before it is sent. That log is retained and auditable. Byte for byte. Question by question.
Our data stays where we say it stays.
What you can sayAn architecture constraint, not a policy
Data residency is enforced at the database level — not as a policy, as an architecture constraint. You choose where. It stays there. A write to a database placed outside its declared region is refused, and nothing is stored.
Access is controlled at the column level.
What you can sayNo separate system to manage it
Sensitive fields can be fenced — visible to some users, invisible to others — without building a separate system to manage it.
Every action is in the audit log.
What you can sayNot sampled — every act
Every read, write, query, and AI interaction is recorded. Not sampled — every act. Your compliance team can pull the full history, and export it as xlsx or csv.
We can show finance exactly what we spent and why.
What you can sayThe invoice reconciles to the meter
Every charge is a dated record, retained 400 days, that adds up to the figure on your usage page. The invoice reconciles to the meter. The meter reconciles to the team and service that incurred it.
These six are what the platform itself does. A certification held by the infrastructure underneath is that provider's to state, and is not claimed here. The whole security and compliance page.
The pricing
One number, visible before you spend
Two cents per thousand operations.
A standard database operation is one unit. A search is five to ten. Storage is ten cents per gibibyte per month. Outbound data is nine cents a gibibyte on the card — the same figure whoever the infrastructure underneath belongs to, because nothing here is priced on somebody else's free tier.
For media workloads specifically, the savings are material:
| Workload | Data Oil | Market rate |
|---|---|---|
| One hour of audio transcribed | $0.24 | $0.26 – $1.44 |
| One gibibyte of text translated, into one language | $1,898.40 | $10,700 – $21,500 |
| One thousand pages of documents read | $4.80 | $10.00 |
In US dollars, at each vendor's published list price. Both columns price the same work in full: the translation row is one target language on both sides, and ours is the source read plus the delivered text, not the source read alone.
The translation figure is not a typo. And competing services charge again, in full, for every additional target language — where our source read is charged once however many languages you ask for, and only the delivered text is charged per language. Fifty languages on one gibibyte is $68,556 here against $805,306 at AWS Translate: nearly twelve times less, and the gap widens with every language you add.
Every rate, with its basis · Beside Snowflake, Databricks and the rest
Two ways in
Two ways to get started — neither commits you to anything
Option 1 — free
The Fixed Price Analysis
Send us the last three months of invoices from the vendors currently holding, searching, and answering questions about your data. You receive a fixed monthly price for the same workload on Data Oil, held for twelve months, with the full working shown. It costs nothing and commits you to nothing.
Option 2 — paid
The Paid Proof of Concept
A proof of concept is a decision instrument, not a demo. Four weeks. One workload you define. Goals agreed in writing in week one. A fixed fee — credited back at 10% of your recurring monthly charges for the first year, up to the value of the fee itself. A written report either way, whether you proceed or not. You leave with a decision, not a meeting summary.
Take these with you
The three questions to take to your next leadership meeting
How many vendors currently hold a copy of our data?
Question oneEvery copy is a bill and a door
Every copy is a monthly bill, a credential to manage, and a potential entry point for a breach.
The full answer →Can we show an auditor exactly what left our environment for the AI model?
Question twoByte for byte, before it was sent
Byte for byte. Question by question. Before it was sent.
The full answer →Can finance reconcile the data bill to a meter — and to the team that incurred it — without building a spreadsheet to do it?
Question threeIf the answer is no, that is a governance gap
If the answer is currently no, that is a governance gap your board will eventually notice before you do.
The full answer →
Data Oil
Every kind of data. One source of truth.
A fixed price from your invoices costs nothing and takes an email. A proof of concept takes four weeks and ends in a written decision.