Secure by design
Least-privilege access and controlled data boundaries.
Security & data handling
How these systems are built to be trusted
Written plainly. These are the operating principles behind HIS engagements and this platform, not certification claims.
Trust principles
Four commitments that shape how systems are designed and delivered.
Human governed
Sensitive actions require explicit authorisation.
Auditable
Actions, tools, approvals and outputs are traceable.
Transparent
Demonstration results are never presented as production outcomes.
Human oversight model
Autonomy is graded by risk. The same model is enforced in the agent registry, not just described here.
Low risk
Autonomous within defined boundaries.
Medium risk
Escalation based on policy or confidence.
High risk
Human approval required.
Critical
Human approval always required.
High-impact actions, including payments, external messages and destructive changes, are designed to pause for human approval before they execute.
Data handling
What happens to your data during an engagement.
Your data stays in scope
Workflows are designed to process the data they were scoped for, in the environment agreed in the engagement. There is no silent reuse of client data to train shared models.
Least-privilege access
Services and agents get the minimum access they need: scoped credentials, per-environment keys and row-level access rules on client-facing data.
You can host it
Processing services are built to run in your own cloud account or infrastructure where required, with the website acting only as an interface layer.
Deletion on request
Uploaded requirement files and engagement data can be removed on request, subject to the retention terms in the Privacy Policy.
Auditability
Systems you cannot inspect are systems you cannot trust.
Every run is recorded
Agent runs carry an execution trail: inputs, tools called, approvals taken and outputs produced, retained for review.
Failures are visible
Failed, paused and held runs are shown in the queue rather than retried silently, so problems surface instead of compounding.
Honest status labels
Demonstration and sample-data results are labelled as such across the site and are never presented as production outcomes.
Responsible disclosure
Found a problem? Tell us directly.
Security concerns can be reported to safdar404@gmail.com. Please include enough detail to reproduce the issue and avoid accessing data that is not yours. Legal details live in the legal centre.
Contact usReady to scope your AI initiative?
Share your objective and constraints. A specialist reviews every request before any recommendation is issued.