P402 Meter
Real Estate · Tenant ScreeningLive demo
> _ P402 METER · REAL ESTATE · TENANT APPLICATION SCREENING

Every applicant
screened.
Every cent visible.

Property managers process thousands of applications. The AI that reads them costs roughly $0.02 to $0.05 per applicant. P402 makes that cost visible, governed by fair-housing rules, and settled onchain: one Tempo tx per application batch.

Synthetic applicantsNo real PIIHuman final decision requiredHUD fair-housing audit posture

Human final decision required. P402 does not approve or deny tenants. The fraud score shown is illustrative; the leasing decision rests with the property manager and follows fair-housing rules.

View dashboard proofSee evidenceRun screening demoAll industries
01The Problem

Tenant screening is high-volume, document-heavy, and legally adversarial.

Legacy
Operating cost per applicant in the legacy workflow
Background check fees plus human review time for income verification and reference checking add up to a materially higher per-applicant cost than the AI-metered path; the exact figure depends on the property and the service mix.
Real
Document fraud in rental applications is a scaled problem
Falsified bank statements and pay stubs are a real, scaled problem in rental applications. AI document review surfaces likely inconsistencies for a human reviewer; the property manager makes the final call.
$0
Per-applicant cost breakdown in current AI screening tools
Existing AI screening platforms (Snappt, MeasureOne) charge per-seat or per-unit. There is no per-applicant AI cost breakdown. Property managers cannot see what the AI actually did or what it cost.
0
Settlement rails for sub-penny per-applicant billing
If you want to charge $0.03 for an AI-assisted tenant screening, you need a settlement rail that handles $0.03. Stripe's per-transaction floor is well above that level. Tempo settles for under $0.000001 per event.
02The Workflow

Six steps. One applicant packet. Full audit trail.

01
Application intake (multimodal)
Upload an applicant packet: rental application form (PNG of a filled-out form), 2 pay stubs (PDFs), 1 bank statement (PDF), 1 ID document (image). Each document is SHA-256 hashed on ingestion. This is the multimodal intake test.
02
Structured extraction
Gemini Flash reads all four documents multimodally. Extracts: claimed income, employer name, employment duration, bank account holder, deposit history, address history, ID name match. Returns a typed Applicant object with a confidence score per field.
03
Cross-document consistency check
Gemini Pro reasons across all four documents simultaneously: does claimed income match pay stubs? Do bank deposits match claimed income? Does the ID name match the application? Does the address on the bank statement match the rental application? Each check produces a consistency flag.
04
Fraud signal and escalation
If the anomaly score crosses a threshold (inconsistent income, metadata manipulation detected, name mismatch), the application is flagged for escalation. In the current demo, this surfaces the case to the human reviewer with Gemini Pro extended-reasoning analysis attached; a future capability will route the case to a specialist agent under MPP escrow.
05
Per-applicant cost readout
Every screening costs roughly $0.02 to $0.05 in tokens. The ledger shows: extraction cost (Flash), consistency check cost (Pro), and any escalated review cost (Pro). The total is visible per applicant and aggregatable across a property.
06
Decision gate: human in the loop
The property manager reviews the AI assessment and approves or denies. P402 records the decision without recording PII; only a redacted applicant ID is stored onchain. The decision log supports a HUD fair-housing and ECOA audit posture; the property manager makes the final decision and the audit log does not replace fair-housing compliance.
03What This Demo Emphasizes

Volume economics. Fraud signal surfacing. Fair-housing audit posture.

$0.03
Avg per applicant
clean application, no escalation
Per-app
Cost visible per application
legacy workflow operating cost is materially higher
4 docs
Multimodal intake
form, pay stubs, bank stmt, ID
·Multimodal volume economics

Property managers screen thousands of applications per year. At $0.03 per applicant, 1,000 applications cost roughly $30 in AI compute; the operating cost in the legacy workflow is materially higher. The economics are not marginal.

·A2A escrow as a real workflow

The fraud specialist escalation is a future capability: a different agent doing a different job under a different budget envelope. The workflow naturally produces it; the current page surfaces the escalated case for a human reviewer.

·Fair-housing governance

The audit trail is load-bearing because tenant screening decisions are legally adversarial. P402 records what AI did, what data it used, what score it produced, and what the human decided, with consistent criteria applied to every applicant.

·Three applicant scenarios

Clean applicant (passes all checks). Income mismatch (low fraud score, manual review flagged). Likely fraud (forged pay stub triggers escalation; specialist agent under MPP escrow is a future capability). Each scenario exercises a different path through the system.

04Sample Applicant Scenarios

Three scenarios. Three paths through the system.

Clean Applicant
All checks pass
~$0.022
Flash extraction, then Pro consistency check; all fields consistent; no fraud signal; property manager approves

Income matches pay stubs. Bank deposits match claimed income. ID name matches application. Address consistent across all four documents. Confidence score: 0.94.

Income Mismatch
Low fraud, manual review flagged
~$0.031
Flash extraction, then Pro consistency check; income flag; anomaly score 0.42; manual review recommended; property manager reviews

Claimed income $95K. Pay stubs show $72K annualized. Bank deposits inconsistent with either figure. Not clear fraud; could be freelance income or recent job change. Flagged for human review.

Likely Fraud
Specialist escalation triggered
~$0.065
Flash extraction, then Pro consistency check; anomaly score 0.87; escalation surfaced to reviewer (specialist agent under MPP escrow is a future capability); deliverable hash on Tempo; property manager denies

Pay stub metadata shows creation date after claimed employment period. Bank statement deposit pattern inconsistent with claimed pay schedule. Gemini Pro forensic analysis confirms document manipulation indicators.

05Unit Economics

Scale is where the economics become undeniable.

VolumeManual screening costP402 AI screening cost
100 apps / monthLegacy: materially higher~$3
500 apps / monthLegacy: materially higher~$15
2,000 apps / monthLegacy: materially higher~$60
10,000 apps / monthLegacy: materially higher~$300

Cost includes Gemini Flash extraction, Gemini Pro consistency check, and Tempo settlement. Escalated fraud cases add ~$0.04 per flagged application. Assumes 10% fraud escalation rate.

See it live

The demo is the proof.

Three synthetic applicant scenarios (clean, mismatch, and likely fraud). Multimodal intake across application form, pay stubs, bank statement, and ID. Cross-document consistency check. Fraud score with escalation threshold. Real USDC.e settlement on Tempo. HUD fair-housing audit posture on every decision; the property manager makes the final call.

Run screening demoVerify on TempoAll Industries
P402 Meter · Real Estate · Tempo Mainnet · MPPTempo × MPP × Gemini 3.1