Skip to content
Public verifier

See an AI decision prove itself.

Pick a real scenario below. Watch what the agent considered, what it chose, and what we'd show a regulator or a judge if anyone challenged it. Every example is anchored on Base Sepolia right now — clickable, checkable, can't be edited.

CareGrid · scenario №03

Triage — verified live

Anchored on Base Sepolia · Apr 26, 2026, 6:36 AM · tx 0x2235d540…

✓ all 6 checks passOpen on easscan ↗
What CareGrid actually got
< 4 min
physician co-sign latency

Attending agreed with the AI route before the patient hung up.

−38%
avoidable ER admissions

The AI escalates only the cases that need it — no defensive routing.

Zero
unexplained triage decisions

Every routing decision carries the signals it weighed and the path it ruled out.

Scenario No.03
CareGrid · telehealth triage agent · acute case

Triage AI routed a chest-pain case from video to in-person ER.

Patient M, 47. Self-reported BP 152/96. Onset 30 min, dull pressure radiating to left arm. AI weighed three care tiers.

What the AI considered
  • Self-care advisory + 24h follow-up (rejected: cardiac signal too acute).
  • Telehealth video within 30 min (rejected: video-only insufficient for chest pain).
  • Refer to in-person ER, arrange transport.
What it chose

ER referral. Attending physician co-signed within 4 minutes; ride dispatched.

Why this matters

A miss here ends careers. The receipt shows exactly which signals the AI weighed, which routes it ruled out, and the second physician who agreed.

Who can ask, and what they ask

The same record satisfies very different questioners. Each gets the same answer — and none of them have to trust the other party's word for it.

The patient

Why was I sent to the ER instead of a video visit?

The payer

Was this routing medically necessary?

Hospital QI team

Are our triage decisions defensible across cases?

What this proves — in plain English
This decision happened.

Recorded on Base Sepolia at April 26, 2026 at 3:36 PM KST. The transaction is permanent and publicly visible — try editing it.

These were the inputs.

Every prompt, every retrieved context, every alternative the AI considered — fingerprinted into the same Merkle tree. Change one byte and the entire root invalidates.

This was the chosen output.

Not just what shipped — what was rejected, scored, and overridden. A future investigator can see the full reasoning, not just the final answer.

A neutral third party witnessed it.

trace.ai anchored the root with 0x4eC5A687…; the operator signed the payload with 0x1e6d0f5e…. Neither party can produce this evidence alone.

tx · 0x2235d5406331305e8ecd5a
Run a simulation

Watch a decision earn its receipt — across 7 industries.

Pick a scenario. Press play. Each step shows a different artifact — the ticket, the score bars, the signature, the Merkle tree, the block, the verifiers — so you see the proof being built. Same protocol works for e-commerce, finance, healthcare, HR, insurance, and legal.

Currently simulating
CS bot processes a refund
01 · Agent receives ticket
02 · AI considers options
03 · Operator signs the record
04 · Hash joins next batch
05 · Anchored on Base Sepolia
06 · Customer · bank · auditor verify
Step 1

Incoming event

The agent receives a real-world signal.

Gorgias#482910
Refund for melatonin order
From: sleepy@example.com

Hi! I bought your 3mg melatonin 9 days ago. Tried it for a week — barely helped me sleep. Could I please get a refund? Order SO-87234.

Captured before the AI reads a token. Every byte will be hashed.
Next

Want to put your own AI agent on the record?

One npm install. One SDK call. Every decision gets a permanent, independently verifiable receipt.

See the quickstart →How it works