Buyer evidence pathScore, inspect, package, request.
  1. Score readiness firstPRISMFind the first deployment gap before a call.
  2. Inspect proofPHANTOMReplay the governed approval boundary.
  3. Verify evidenceEvidence roomChoose role and workflow proof.
  4. Build passportPassportPackage readiness signals into one artifact.
  5. Send ATLAS intakeRequestCarry context into founder review.

Public proof console

PHANTOM proof console

Inspect a synthetic governed commerce run now. Private runtime walkthroughs are issued separately after founder review.

Inspect AgentOps console →Request private walkthrough →
Public landing: VercelAgent runtime: private APIData: synthetic mock onlyLive writes: disabled

Public synthetic proof

Replay the approval boundary without private credentials.

The sample uses mock commerce data and shows the evidence a buyer committee needs to inspect: proposed action, risk class, human approval gate, audit trail, CLEAR score, and AgentOps handoff.

Open evidence room

The proof output below is already loaded with synthetic data.

Proof reading guide

Read PHANTOM through the same buyer questions procurement will ask.

Keep PHANTOM as the proof object, then choose the smallest next surface for the stakeholder in front of you: trust evidence, readiness, passport, or founder-reviewed intake.

Evidence roomScore readinessBuild passportFounder review

90-second governed replay

Watch PHANTOM move from READ_ONLY analysis to a held FINANCIAL boundary.

This is the buyer-readable version of the audit trail below. Step through the run, then switch buyer lenses to see why the same evidence matters to operations, data, risk, and the sponsor.

Approval gate

Human approval boundary held

The workflow pauses at the FINANCIAL boundary. A human rep must decide before any consequential commerce action can happen.

Evidence
Approval event, run status, persistent audit, and AgentOps handoff.
Buyer signal
Risk, legal, and operations can verify who is accountable for the decision.
Boundary
No order submitted. The public sample cannot approve, submit, update, refund, discount, or write.

Buyer committee proof lens

Same run, different committee question.

Operations

Will this improve the rep workflow?

The visit packet compresses account context, SKU urgency, promo gaps, and next-best actions into one reviewable operating view.

  • Rep-visible visit brief
  • Suggested order lines with urgency
  • Next-best actions before the approval decision
Carry this proof into founder review

Private runtime

Run a guarded visit package with issued credentials

This control is for qualified demos. It uses synthetic accounts and blocks all consequential writes behind the PHANTOM approval boundary.

Advanced runtime settings

Runtime URL stays browser-side for this private console. Production prospect pages should use the readiness route, not direct runtime controls.

Private runtime locked. Use the public sample above or paste the issued demo token.

Governance evidence

PHANTOM VSA

SYNTHETIC_SAMPLE
Risk maxFINANCIAL
ThreadPUBLIC-P…LE-001
CLEAR1.00
Boundarymock only
Live-system writesdisabled
Persistent auditverified

Visit package

El Corte Market - Diagonal

Synthetic account package: two fast-moving SKUs are below target stock, promo visibility is below the planogram threshold, and the proposed order remains held for rep approval.

Planogram 72%Order €1,248.752 proposed lines

Suggested order

SKUUrgencyStockDaysCases
Citrus Sparkling 12x330mlhigh1424
Protein Bar Cocoa 24ctmedium2243

Promo compliance

  • Cold cabinet visibilityDisplay share is below agreed planogram threshold.
  • End-cap launchTwo required SKUs are not visible in the synthetic store check.

Next best actions

  • Ask the rep to confirm display correction before approving the order.
  • Prioritize Citrus Sparkling replenishment because stock-out risk is highest.
  • Log the promo gap for the account manager before the next visit.

Buyer proof packet

Proof a committee can inspect.

A compact artifact for the first call: proposed action, approval boundary, audit evidence, CLEAR baseline, and AgentOps persistence.

PENDING_REP_REVIEWPUBLIC-PHANTOM-SAMPLE-001
Proposed action2 order lines€1,248.75; not submitted.
Approval boundaryFINANCIALHuman review before consequential action.
Evidence density5 audit events1 approval event(s).
MeasurementCLEAR 1.00Planogram 72%.
  • Rep-visible recommendation, not autonomous order submission.
  • Risk class and public execution boundary are explicit.
  • Audit rows and AgentOps evidence are ready for founder review.
Send proof context to founder review

Conversion path

Governed proof delivered.

PHANTOM just ran a governed visit package with audit trail, FINANCIAL boundary, CLEAR scorecard, and synthetic commerce data.

Book diagnostic sprint

Audit trail

01account_readREAD_ONLYAccount context loaded
02sku_velocity_readREAD_ONLYSKU velocity checked
03planogram_readREAD_ONLYPlanogram compliance calculated
04order_proposeREAD_ONLYOrder proposal generated
05approval_eventFINANCIALHuman approval boundary held

CLEAR scorecard

Cost1.00< EUR 0.80
Latency1.00< 30s
Efficiency1.00< 12 tool calls
Accuracy1.00>= 0.85
Reliability1.00>= 0.95

AgentOps persistence

Durable rows5
Approval events1
Financial boundary1
Thread matchtrue

This run has durable Supabase audit evidence for AgentOps reporting.

Inspect this run in AgentOps