KAS · in build
Your agents do the work. You set the limits.
AI agents can do real work for your firm.
Until now, you could not fully trust them to stop.
Send a quoteAllowed
Pay $40,000Blocked
They can send a quote, pay a supplier, or book a meeting.
KAS stops an agent from doing anything you have not allowed.
Qypu stands behind that.
The picture and the list on this page are examples. They are not a live record. KAS is still in build.
01
Prevent
You write a list of what each agent can do. KAS checks that list before a job starts.
The list names the jobs, the spend, and the dates. The job starts only after a yes. If KAS cannot check, the job does not start.
02
Guarantee
KAS guarantees that an agent does not do a job you have not allowed. Qypu stands behind that guarantee.
03
See
You can see what your agents have done, whenever you want to look.
What your agents did today
An example. Not a live record.
- Agent sent the quoteAllowed
- Agent tried to pay a supplier $40,000Blockedyou did not allow it
- Agent booked Thursday's meetingAllowed
- Agent tried to email every contactBlockedyour permission ran out
- Agent matched 3 paymentsAllowed
- Agent tried to delete last month's invoicesBlockedyou did not allow it
- A colleague checked the day's listChecked
KAS writes down each yes, each no, and each pause. No one changes an old line. A second person can check that record. That person does not allow a job, and does not block one.
What we can say today
We hold no certificate yet.
Security
- SOC 2 Type IIIndependent audit in progress
- Cyber Essentials (UK)Certification in progress
- Cyber Essentials (Singapore)Certification in progress
- CyberSecure CanadaCertification in progress
- Essential Eight (Australia)Alignment in progress
AI governance
- EU AI ActReadiness review in progress
- IMDA agentic AI frameworkBuilt to follow it. Review in progress.
Privacy
- PrivacyCompliance review in progress
Statuses are current as of October 2026.
Trust centre- SOC 2 Type II. Independent audit in progress
- Cyber Essentials (UK). Certification in progress
- Cyber Essentials mark (Singapore CSA). Certification in progress
- CyberSecure Canada. Certification in progress
- Australia Essential Eight. Alignment in progress
- EU AI Act. Readiness review in progress
- Singapore IMDA Model AI Governance Framework for Agentic AI. Built to follow it. Review in progress.
- Privacy. Compliance review in progress Singapore PDPA, Canada PIPEDA, Quebec Law 25, the Australia Privacy Act, and the NZ Privacy Act.