Your AI makes the decision.
We help prove a human approved it.

HumanGate™ intercepts AI decisions before they execute and routes them to your team for review. Every verdict produces a signed Certificate of Oversight™ that anyone can verify independently.

Book a Demo How it works
Certificate of Oversight™
SIGNATURE VALID S. Chen · Approved
CERTIFICATE OF OVERSIGHT™ cert-a8e2f1b3
OrganizationBrownie’s Credit Union
DecisionLoan Approval · $45,000
AI AgentLoanBot v2.1
Risk Score72 / 100
VerdictAPPROVED
ReviewerS. Chen, Senior Analyst
Duration2m 14s
SignatureRSA-2048
Verifying signature…
PENDING
The problem

Your AI agent denied a loan.
A regulator asks for evidence of human oversight.
What do you show them?

TODAY
No record of who reviewed it
No documented rationale
No verifiable timestamp
Evidence can be altered after the fact
No way for a third party to confirm it happened
Brownie’s Credit Union reviewed this loan denial in a Slack thread. The regulator sees a screenshot.
WITH HUMANGATE
S. Chen, Senior Analyst — named and recorded
Written rationale attached to the decision
2026-03-04 14:22 UTC — immutable timestamp
Digitally signed — cannot be altered
Verifiable by anyone with the public key
The regulator verifies the certificate independently. No access to your systems required.
How it works

One API call. Human review. Signed certificate.

Every high-stakes decision is held for review before it executes.
The output is a signed certificate — tamper-evident and publicly verifiable.
01
Submit

Your AI agent sends each decision via API before acting on it. HumanGate evaluates risk and assigns a review priority.

02
Review

A designated reviewer evaluates the decision with full context: the AI’s reasoning, risk assessment, and the data behind it.

03
Certify

HumanGate generates a Certificate of Oversight™ signed with your organization’s keys. The verdict, rationale, and reviewer are permanently recorded.

04
Return

Your system receives the verdict via webhook. The certificate is independently verifiable by anyone — regulators, auditors, or counterparties.

Approved

The decision met your review criteria. It executes with a signed, auditable record attached.

Rejected

The decision did not pass your review. It is returned to the agent with documented rationale. The rejection is also certified.

Runs on your infrastructure.
Your keys. Your data.

HumanGate deploys behind your firewall. Your database, your signing keys, your network perimeter. We never see your decisions, your reviews, or your certificates.

The certificate

What an auditor sees
when they verify a Certificate of Oversight™.

Every certificate can be checked by anyone. No login, no API key, no access to your environment. Just the certificate and your public key. The math does the rest.

  • RSA-2048 digital signature (PKCS1v15 + SHA-256)
  • PDF integrity hash verified independently
  • QR code links to public verification page
Verification Result
Digital signatureVALID
PDF integrityINTACT
Certificate IDcert-a8e2f1b3
ReviewerS. Chen
VerdictAPPROVED
Timestamp2026-03-04 14:22 UTC
Issuing keyBrownie’s CU · RSA-2048
FAQ

Questions we hear most

How does it deploy?
Three commands with Python, or docker-compose up if you prefer containers. Air-gapped operation supported.
Does HumanGate see my data?
No. It’s software you run, not a service you send data to.
Who does the reviews?
Your team. Compliance officers, risk analysts, operations staff — anyone you designate. HumanGate provides the system. Your people provide the judgment.
Does HumanGate take on liability?
No. HumanGate documents that oversight occurred. Decision authority and liability remain with your organization.
How long does integration take?
One endpoint. One POST request before execution. Verdict returned via webhook. Most teams integrate in under a day.

See HumanGate in action.

Your personal demo will be ready within 24 hours.

You’re in.

We’re setting up your personal demo. You’ll receive login credentials by email shortly.

Read: Why HumanGate → Read: Product Spec → Enterprise Brief →