Your data is never used to train AI models
Documents, claims, decisions and recordings are processed only to produce output for your own workspace. Nothing you upload is used to train, fine-tune or improve any model — ours or a vendor's.
Trust & data protection
Strategy work only happens where the material is safe.
Venture data is the most sensitive material a company holds. These are the commitments we operate under — written in plain language, without exposing the internals of the system.
Closed-loop AI processing
Analysis runs inside our controlled pipeline against a fixed set of approved models, invoked server-side with a scoped request and no retention beyond producing your result. There are no open chat integrations, plug-ins or third parties that can pull your venture content out of the loop.
Strict tenant isolation
Every record belongs to exactly one organisation, and access is enforced at the database layer rather than in the interface. A user without membership cannot read a row, even through a direct query.
A private vault for source material
Uploads live in private storage that is not publicly addressable. Access is granted per organisation member and served through short-lived, expiring links.
Nothing is shared unless you share it
There is no cross-venture pooling, no benchmark that exposes your figures, and no default sharing. External access exists only when you create a named, expiring data-room invitation — watermarked, section-scoped, with downloads off by default.
AI proposes, people approve
No AI output changes your venture state on its own. Each proposal carries its model, inputs, confidence and reasoning, and waits for an authorised person to accept, edit or reject it.
Everything material is logged
Strategic changes, approvals, subscription events and every external data-room open, view and download are written to an append-only record you can review.
Your data stays yours
You can export your venture state and source documents, and you can request deletion of a workspace and its vault. Meeting recordings can be discarded once the minutes are approved.
Questions we get asked
Straight answers
- Which AI models are used?
- A fixed set of enterprise-grade hosted models, invoked server-side under a data-processing agreement that prohibits training on submitted content. Models are selected per task and can be reviewed with your team on request.
- Do your staff read our venture data?
- No routine access. Support access requires a specific request from you and is logged.
- Where is data stored?
- In managed EU-region infrastructure with encryption in transit and at rest, and daily backups.
- Can we get a security review or DPA?
- Yes. Team and Pro customers can request a security overview, sub-processor list and a signed data-processing agreement.