The questions asked before anyone signs in.

No, and there is nothing to approve or request access to.

EnableDat works from what you know about the business, not from your data. A session is questions and answers: which decision needs to be right, what goes wrong when the data behind it is wrong, what your team already knows.

No. What the work needs is not in a file.

An export shows you columns. It cannot tell you which decision depends on them. That sits with you. In one session we ran, the highest-priority data element was not a field at all: it sat in free text in a notes field. A file would have shown everything except the one thing that mattered.

No. If we took your data we would protect it: everything stored sits in the EU and is never used to train models.

We do not take it because data cannot answer what EnableDat asks: which decision depends on it, and what goes wrong when it is wrong. That is a design decision about where the value is, not a limitation we are working around.

What you type and what each session produces. Application data in the EU via Supabase, Dublin. Hosting and processing in the EU via Vercel, Stockholm. Model inference via the Anthropic API, which does not use API traffic to train its models.

You can export or delete at any time. Full detail at enabledat.com/security.

Ask ChatGPT for data quality rules and you get rules in thirty seconds. That is the problem. It will accept whatever you tell it, invent a field your system does not have, cite a regulation it has not read, and write the rule before anyone has asked what decision the data is for. It looks finished. It protects nothing. And the next person who asks gets a different answer.

EnableDat is built to not do that. It will not write a data quality rule until the business decision behind it is clear. It pushes back when an answer does not hold. It only uses the fields you named. It says "I don't know" instead of guessing, and it stops to ask a person when a person should decide. That behaviour is set by over 200 written standards and rules, drawn from 200+ real business rules across 11 industries, and every change is tested against them. So two stewards in two departments get output that reads the same, and it still does next year, after the model has changed.

You could build the chat by Friday. The years of knowing what it must not do are what you would be rebuilding.

The logic is the easy half. Any model will write you the textbook: the dimensions, a data quality rule for each. Every company gets the same text.

The hard half is knowing what to ask a steward first, and when the answer does not hold. Try it: ask your model for the data quality rules behind "invoices must be paid within 30 days". Five rules in ten seconds. Now ask which decision they protect and which field it reads. It does not know. It never asked. EnableDat asks, and will not write a rule until it knows. That judgement is not in any book. It came from years in the meetings where rules get written and ignored, and from watching the engine fail with real stewards until it stopped.

Then comes the cost nobody budgets for: keeping it right as models change, for years. That is our whole job. For you it is one more tool to maintain, outside what your team is there to do.

Everything EnableDat produces is a plain document you own. When your own AI is ready, it starts from those.

Technically yes. The model is chosen per step from a registry, not hardcoded.

But every model behaves differently, so a change means a full test round per module before it is locked. We have done it several times, and each time found behaviour that only shows up in real sessions.

No. The method (the steps, the guardrails, the review standards) is what makes two sessions comparable, whoever ran them. That part is the product.

It does not look at your data, and it will not tell you whether your data is correct. It tells you where the risk sits, based on what you know about it.

It does not replace a data catalogue or a monitoring platform. It decides what is worth putting into one.

And it does not decide for you. You are the source of truth; EnableDat is the method.

Data stewards are the primary audience. It is also used by data governance leads, data quality managers and data domain owners: anyone who has to turn a governance question into something concrete.

No. It structures the work; the steward brings the context, the domain knowledge and the judgement. The result is theirs to check and to own.

Because the role means translating broad governance ambitions into concrete work, usually with a fraction of their time and no expert on call. That takes structure, a way to prioritise, and a result others can read. That is what a session gives them.