A reasoning companion for data stewardship. Find what matters, settle the language, ship the rules — and keep what your team decides. No access to your data required.
A reasoning companion for data stewardship. Get clarity on what data matters to your business and what to do about it.
Takes less than 2 minutes.
Signing you in…
Quick plain-English summary. Full details on the data handling page linked below.
Scroll to the end of the document to continue.
Optional — help us improve the product
Your account information, session content, and generated outputs are stored in Supabase, hosted in the EU (Dublin, Ireland). Serverless functions that process your requests run in Vercel's Stockholm region.
Three sub-processors, all with published security certifications:
Use enabledat for any legitimate data governance work inside your own organisation. The following are not permitted: building a competing product from session outputs, using session content to train or evaluate AI models, systematically extracting the platform's workflow or prompts, sharing access credentials with third parties, or bypassing the platform's controls. The artefacts your sessions produce are yours; the workflow that produced them is not. These restrictions survive account termination.
We monitor session activity in aggregate (counts, completion rates, modules touched, cadence between turns) for security, abuse prevention, and product improvement. We do not read session content as a matter of course. Access to individual session content is restricted to named administrative personnel and used only when investigating a specific suspected breach of the acceptable-use terms, a security incident, or with your explicit consent (for example, when you submit feedback that references a specific session).
Under GDPR you have the right to access, correct, export, and delete your data at any time.
Consent to our terms is required to use the product. Optional consents (product improvement, case studies, product emails) are truly optional and can be changed at any time in settings. Withdrawing consent does not affect the lawfulness of processing before the withdrawal.
All data is encrypted in transit and at rest. Row-level security on every table ensures that one user cannot read another user's data through the application. Administrative access is restricted to named personnel with two-factor authentication.
For data requests, security reports, or any privacy questions, .
Choose the language for the app. Your choice is saved and the URL changes accordingly.
Change these any time. Changes are logged.
Email reminders for pending follow-up work.
Download a copy of everything we hold about you — profile, sessions, generated outputs, feedback, consent history. Returns as a JSON file (GDPR Art. 15 + 20).
Deactivates your account immediately and permanently erases everything seven days later. You can cancel within the seven-day window by signing back in. Knowledge records you contributed to your organisation are anonymised and retained for the organisation.
Your account will be deactivated immediately and permanently erased after a seven-day grace window. You can cancel within that window by signing back in.
To confirm, type DELETE below:
Privacy questions, data requests, security reports, or anything else. We typically reply within a working day.
We've made changes to our terms of service and privacy policy. Please review and accept to continue.
Choose a strong password. You'll use it every time you sign in.
Must be at least 8 characters.
Which modules to enable
Welcome message (optional)
0 users selected.
Cost is a client-side estimate from lib/cost-tracking.js. Anthropic Console remains authoritative.
% of each week's new signups who returned in the following weeks. Blank cell = the window hasn't fully elapsed yet.
Cost figures are estimates from lib/cost-tracking.js. Anthropic Console remains authoritative.
Heuristic signals that suggest a user may be mapping the system rather than doing real governance work. A high score means worth a look, not definitely scraping — investigate before acting. Window: last 30 days. See the Guide tab for how to read these.
Six heuristics, one composite score per user. None of them prove someone is mapping the system; together they tell you who's worth a closer look. This guide explains how to use the table without over-reacting to it.
Each user gets one point per flag they trigger. Flags are independent — they don't double-count the same behaviour. The maximum is 6, but in practice you'll mostly see 0–3.
Why this might be mapping: the platform has four modules. Someone trying to understand the product as a whole hits all of them quickly. Real users almost always have one or two modules they live in.
Why it might be a false positive: a curious tester who you've just given access to all four. A data governance lead exploring the whole offer. Both legitimate.
Why this might be mapping: "let me see what each one does." Concentration of breadth right after signup, before depth has happened anywhere.
Why it might be a false positive: the onboarding routing flow shows people what's available. Some users will dip into multiple modules before committing to one. Look at completion in the modules they touched — real users finish at least one session somewhere.
Why this might be mapping: opening a session to inspect what step 1 asks, then abandoning before step 2 reveals the workflow shape. Common pattern for someone trying to learn the system, not use it.
Why it might be a false positive: a real user who creates draft sessions and decides not to pursue them. Frustration with the tool on a confusing step. Check the abandoned sessions — if they all stopped at step 1 or 2, that's the mapping pattern. If they stopped at varied steps, it's frustration.
Why this might be mapping: minimum input to advance the engine, no real reasoning. Sessions ended early because the user got what they wanted from looking at step 1 or 2.
Why it might be a false positive: short, focused sessions with clean inputs (which the platform actually rewards). The Rule Engine in particular can be efficient at 8–12 turns; under 5 is unusually short. If their median is 4 and the rules look real, it's not mapping.
Why this might be mapping: a brand new account creating a burst of sessions. Burst-then-quiet is the rhythm of someone who came to look around, not to use the tool.
Why it might be a false positive: a genuinely excited new user who you've just activated. Their first day of activity is naturally heavier than steady-state. Check what happened on day 2 — if it dropped to zero and stayed there, mapping. If sessions continued, real adoption.
Why this might be mapping: faster than a human typing a real answer. Consistent with copy-paste from a script or another window. Also consistent with someone running an LLM client that auto-responds to your engine.
Why it might be a false positive: a user who pre-wrote their answers in a notes file and is pasting them in. That's actually a sign they take the work seriously. Read the content of those turns before assuming the worst.
Last reviewed: 2026-05-21. Update this guide if you change thresholds in lib/handlers/admin/behavior-signals.js.
Identify critical elements, risks, and contested concepts in your domain.
Turn a business rule into structured, justified data quality rules.
Reach an agreed, unambiguous definition for a contested business term. Astrakan method.
Turn a methodology gap into a structured starting point for the procedure your team has been working without.
The platform is in early access. Request access and we will get in touch within two business days to set you up.
Thanks. We will get in touch within two business days at the email on your account.
Pick a module above to start your first session.
No sessions yet.
This session surfaced items that need follow-up work. Tackle them now, save for later, or dismiss.
Definition structured using the Astrakan method for business concept modelling.
A starter scaffold — refine with your data lead, compliance, and any relevant business owner before signing off.
Here's what came up during this session that's worth following up on. Pick one up now, save it for later, or dismiss what doesn't apply.
Nothing else came up worth following up on. Nice work.
Thanks — noted.
This text will be appended to lib/prompts.js and deployed automatically.