Skip to content

Trust brief · 1 min read

Responsible startup intelligence

Keep observations, interpretations, permissions, and recommendations distinct.

Distinguish fact from inference

A recorded observation describes what happened. A founder claim explains a belief. An inference connects observations to a possible explanation. A recommendation proposes an action. Labeling these distinctions helps readers challenge a conclusion without erasing the original evidence.

Make recommendations inspectable

A next-action suggestion should identify its source and explain its limitations. If a recommendation cannot be traced to evidence or a stated assumption, treat it as an idea for review rather than an instruction. A human remains responsible for consequential business decisions.

Treat access as a boundary

Role labels are not authentication. Private data needs real sessions and workspace-scoped authorization, including every server-side read and mutation. A read-only interface must also use read-only data adapters; hiding a save button alone does not secure an endpoint.

Describe the actual release

In this release, the public demo uses fictional records and does not accept edits or AI prompts. Production accounts, lead collection, billing, and scheduling remain gated. The trust page explains those boundaries; the Terms and Privacy pages clearly state their pending approval instead of presenting invented legal coverage.

From reading to doing

See the loop in context.

Explore a fictional founder workspace. No account or personal information required.

Explore the read-only demo