Public judge demo · Read-only technical proof Technical proof — not the product interface. This page shows the verified seven-agent architecture, deterministic tools, provenance, replay and reference-run evidence. Live Gemini execution is intentionally unavailable to anonymous visitors.
← Open AOO product View verified run
A
Autonomous Opportunity Operator
Google Cloud provenance loading…
Google ADK · Gemini · Deterministic Authority

From messy opportunity to evidence-backed decision.

A seven-agent Google ADK team reads the source, verifies the facts, checks constraints, remembers prior failures, tests the economics, and stops before consequential external action without human approval.

Not a chatbot. It completes an auditable workflow; you keep authority.
Verified Reference Run
Real cloud execution · Aug 14
FINAL OUTCOME
DECISION
REQUIRED
Human approval is required before registration, cloud resource creation or external submission.
The system completed the real seven-agent workflow in 94.3s after previously hitting a 300s timeout—then stopped exactly where human authority begins.
47,306 source characters verified
7 specialist agents completed the job
4 deterministic checks remained authoritative
0 consequential actions without approval
Autonomous agent team
Seven specialists hand off evidence, not chat
7 agents
Discovery
verified execution
Primary Source Verification
verified execution
Deterministic Hard Gate
verified execution
Investigation
verified execution
Failure Memory
verified execution
Economic Evidence
verified execution
Final Adjudication
verified execution
Workflow trace
Every stage completed and remained auditable
1
PRIMARY SOURCE VERIFICATION ADK evidence observed
✓
2
DETERMINISTIC HARD GATE ADK evidence observed
✓
3
INVESTIGATION ADK evidence observed
✓
4
FAILURE MEMORY ADK evidence observed
✓
5
ECONOMIC EVIDENCE ADK evidence observed
✓
6
FINAL ADJUDICATION ADK evidence observed
✓
Deterministic authority layer
The model cannot overrule these results
◆ Eligibility · Capital · Deadline Gate authoritative
◆ Failure Memory Similarity Check authoritative
◆ Unit Economics authoritative
◆ Final Evidence & Safety Adjudication authoritative
Human authority boundary
DECISION REQUIRED
DECISION_REQUIRED
  • Registration requires your approvalREGISTRATION_APPROVAL_REQUIRED
  • Cloud resource creation requires your approvalCLOUD_RESOURCE_CREATION_APPROVAL_REQUIRED
  • External submission requires your approvalEXTERNAL_SUBMISSION_APPROVAL_REQUIRED
Proof, not claims
Primary-source and durable-runtime evidence
Primary source
https://allthingsagentichackathon.devpost.com/rules
Evidence fingerprint · SHA-256
401079adb7c18cf5bfe1c70a081e82628dd81b9c950ac8ecaab7a3d32c355285
47,306 verified text characters
17 normalized ADK events
<1s durable replay
Replay guarantee
Same source + same decision profile → 0 additional ADK workflows.
Live Analysis

Give the team a real opportunity.

Nothing runs automatically. Preflight checks only the source. The seven-agent Gemini workflow starts only when you explicitly press the live-analysis button.

What the team will do Source preflight is model-free. The Google ADK workflow starts only after your explicit click.
01
Read & verify Capture the primary source, verify facts and preserve provenance.
02
Test & remember Apply deterministic gates, investigate execution, check failure memory and economics.
03
Decide & stop safely Produce an evidence-backed disposition and stop before any action requiring human authority.
Final disposition
—
— agents observed
— tools called
— coordinator stages
— replayed
Evidence