Public synthetic sandbox — no real customer data, no account, no upload, no live AI execution.

Operating model

A governed loop with a human always in charge

SOPCam AI connects the people, the assistant and the record into one accountable workflow. The public sandbox walks exactly this model, on invented data.

Frontline worker

Opens a clear, step-by-step checklist on their phone — in their language — and records completion. No dashboards, no jargon.

Manager / reviewer

Captures the task, reviews the AI draft, edits it, and approves. Nothing reaches a worker without this approval.

AI assistant

Proposes a first-draft SOP from the captured task and flags where it is uncertain. It never publishes and never overrides a human.

SOP library

Holds approved, versioned procedures per site and task, so the current standard is always the one workers see.

Evidence & audit trail

Records who did what and when — capture, draft, edits, approval, publication, and worker completion — as traceable events.

Multilingual worker view

The same approved SOP, presented with worker-language labels (English + Hindi shown in this demo) so frontline teams actually understand it.

The 10 stages the sandbox walks

Eight core stages, then two that show what happens after a procedure is live.

  1. Capture Where the knowledge comes from
  2. AI draft A first draft, clearly marked
  3. Review A person corrects it
  4. Approval An explicit human decision
  5. Publish Release the approved version
  6. Worker view What the frontline sees
  7. Training Understanding, not just reading
  8. Governance Versions, evidence, audit
  9. Multilingual (optional) The same version, another language
  10. Change impact (optional) One edit, traced forward

Governance controls

The rules that keep the loop accountable.

  • AI drafts; a human reviews and approves before anything publishes.
  • Approval and publication are separate, explicit actions.
  • Every externally visible step is backed by a traceable event.
  • Confidence and provenance are shown, not hidden.
  • No worker surveillance — the system trains and verifies competency, it does not monitor people.