Agent Work Receipt: Runtime Trust Infrastructure

Prove AI-agent work. Enforce policy. Defend billing.

AI agents are booking procedures, collecting deposits, updating systems of record, resolving tickets, and triggering invoices at scale. Real work. Real money. Real accountability. As soon as agents do real work, you need more than logs.

Agent Work Receipt is the compliance and revenue evidence layer that proves what AI agents did, whether policy was followed, and whether the work is billable. Runtime trust infrastructure for AI agents. One integration connects any AI agent or platform. Evidence-ready in days, not months.

HIPAA FDCPA GLBA PCI DSS TCPA Insurance 42 CFR Part 2 AI Governance GENIUS Act Ready
The Trust Shift

When agents move from answering to doing, the trust problem changes entirely.

An agent that generates a summary is useful. An agent that books a procedure, captures a deposit, updates a patient record, resolves a ticket, or triggers an invoice is performing economically meaningful work.

Once agents do real work, your accuracy question becomes a much harder set of questions. Logs, observability tools, billing platforms, and GRC systems each answer a fragment. None of them answer all of it.

Agent Work Receipt is the missing connective tissue: the proof object that answers all of them in a single durable record that survives billing, compliance, audit, disputes, settlement, and handoff.

From one question
Was the answer accurate?
To ten questions every business now asks
Did the agent actually complete the task?
Was the agent authorized to do it?
Was the system of record actually updated?
Was sensitive data handled correctly?
What tokens, tools, voice minutes, or actions caused the charge?
Is this work actually billable?
Can the buyer dispute it, and what do we show them?
Can finance approve this bill?
Can compliance audit it?
Can another agent or human safely continue from here?

The same root problem, by who you are

Agent Work Receipt solves a different surface of the same underlying gap depending on where you sit in the stack.

Platform founders and CTOs
Your enterprise buyers are asking for evidence you do not ship yet.

Healthcare, insurance, and financial-services deals stall on one question: "Can you give us a per-action evidence package, policy record, and audit export?" The answer today is no. Building it internally takes 18 to 24 months and is not your core roadmap. Agent Work Receipt embeds in your stack so you can close those deals without building compliance infrastructure from scratch.

Agent developers and AI agencies
Your clients want evidence. You do not want to become a compliance company.

Every regulated client engagement creates a new compliance problem: per-client policy packs, evidence retention, billing proof, and audit support. Building it per-client makes smaller engagements unprofitable and larger ones risky. Register your agents once, apply the right policy pack per client, and deliver client-ready work evidence without building the trust layer yourself.

Technical architects
Every new platform creates another evidence-plumbing problem.

You are stitching webhooks, schemas, idempotency logic, evidence retention, and audit exports across multiple agent runtimes, tools, and systems of record. Every new platform or client adds another one-off integration. Agent Work Receipt provides a canonical event envelope, Runtime Gateway, provider adapters, idempotency enforcement, tenant isolation, and stable billing-grade APIs that work across your entire stack.

Regulated operations: healthcare, insurance, finance, legal
Agents are touching patients, claims, payments, and records. You cannot always prove what happened.

A failed appointment write, a bad consent record, a missing audit trail, or a disputed bill creates compliance, revenue, and customer-risk exposure at the same time. When OCR, a state regulator, or a payer asks for per-action evidence, reconstructing it from transcripts, logs, and billing records takes weeks. Agent Work Receipt gives you the per-action record at the moment the action happens, not after the inquiry opens.

AI-BPO and agents-as-a-service providers
Your clients pay for agent work. They are starting to ask for proof it happened.

Invoice disputes, SLA questions, and exception reports consume delivery margin and erode client trust. Rebuilding evidence manually for each client inquiry is not scalable. Managed Work Evidence packages turn your agent activity into client-ready proof: completion records, SLA evidence, acceptance history, exception detail, and dispute defense, without adding headcount.

Revenue operations and finance
Your AI bills are growing. Finance cannot approve what it cannot verify.

Token bills, voice-minute invoices, tool-credit charges, and outcome-based pricing lines are showing up without any connection to the actual work that produced them. Disputes become manual and subjective. AI Bill Proof maps usage charges back to validated Agent Work Receipts, showing which spend produced evidenced work, which lines lack proof, and which are in dispute or awaiting credit.

The Problem

Agent work without proof is
invisible, ungoverned, and indefensible.

A voice agent books a $1,200 procedure. The system-of-record write fails silently. The patient gets billed. They dispute it. Nobody can prove what the agent did. A scribe agent processes 4,000 visits a month, a regulator opens an audit, and logs do not equal an evidence package. Your CFO learns that 18% of "completed AI workflows" were either duplicate-billed or never had the required system-of-record write. The compliance problem and the revenue problem are the same problem.

Without Agent Work Receipt
Silent system-of-record failure
Agent books a $1,200 procedure. The write fails silently. The clinic bills the patient. The patient disputes it. No one can prove what the agent did. The dispute escalates to legal.
Volume-driven audit demand
Your scribe agent processes 4,000 visits a month. A regulator opens an audit. You have logs. Logs are not an evidence package. Your team cannot assemble the required bundle inside a 30-day window.
AI revenue restatement
18% of "completed AI workflows" billed were either duplicate-billed or never had the required system-of-record write. Your CFO learns "billable event" and "defensibly billable event" are different statements.
Class-action discovery
Your platform has two years of call recordings. Discovery demands per-call consent records and per-call policy versions. No per-call policy record exists. The discovery becomes a forensic exercise.
Revenue that cannot be billed
Agent work that cannot be proven cannot be billed. Payers dispute outcomes. Patients dispute charges. Without evidence, you absorb the loss.
With Agent Work Receipt
Event truth: the work happened
Every agent action is captured with a cryptographic identity, timestamps, and system-of-record write confirmation. Each agent has its own ledger. The work happened. You can prove it.
Event Ledger Write Verification Correlation IDs
Policy followed: the decision record exists
Every action is evaluated against your rule packs at runtime. A decision receipt with verdict, reasons, and evidence attaches to the action before it exits the evidence boundary. Policy version is locked to the event.
Policy Engine Decision Receipts Action Interception
Evidence integrity: retained under the correct rule
Transcripts, policy decisions, system-write proofs, and billability state retained encrypted for 6 years under the correct regulatory retention rule. Reason-coded access. Immutable chain of custody.
6-Year Retention PHI Redaction Audit Bundles
Defensible billing: billed, audited, disputed
Every compliant agent action becomes a trusted billable event with provenance. Export earned events to your billing system. Answer audits in minutes. Drill from an invoice line to the decision receipt behind it.
Validated Work Events Billing Export Dispute Support
Signed evidence package: audit answered in minutes
When a regulator opens an audit or a customer files a dispute, Agent Work Receipt assembles a signed evidence package: what happened, which policy applied, what evidence was retained, and why the outcome was permitted. Not months. Minutes.
Signed Packages Policy Lineage Real-time Dashboard
How It Works

From agent action to billable, compliant proof

Before cloud hosting, businesses carried the entire server infrastructure stack themselves. Agent Work Receipt does for AI agents what cloud did for websites: it collapses the fixed cost of doing AI-agent work responsibly into a marginal cost. Register an agent once and every action is captured, policy-checked, audited, retained as evidence, and turned into defensible billable work automatically.

1. Connect your agents

Point your agent's event stream at Agent Work Receipt's ingestion API via mTLS (Hello Voice and internal agents) or HMAC-signed webhooks (any external platform). One integration. Every agent action enters the evidence chain.

mTLS Certificate Tree
Webhook Signature Verification
Platform-Agnostic

2. Enforce policy, validate the work

Every agent action is evaluated against your rule packs (HIPAA, PCI DSS, FDCPA, TCPA, 42 CFR Part 2) in milliseconds. Agent Work Receipt validates required evidence, suppresses duplicates, confirms system-of-record writes, and determines whether the work event is billable.

Real-time Redaction
Action Interception
Consent Validation

3. Retain evidence, emit trusted events

Agent Work Receipt stores the evidence package behind every action: transcripts, policy decisions, system-write proofs, and billability state. Audit-ready retention. Trusted event feed to your billing and analytics systems. Dispute support when customers push back.

Epic (FHIR R4)
Stripe & Salesforce
HubSpot & AWS SES
Live Scenario Northwind AI + Beacon Dermatology

Northwind AI is a voice AI agency. They register the Beacon Dermatology Front Desk Agent in Mercury, connect Vapi, and point the event stream at Mercury's webhook endpoint. From that moment, every after-hours call is captured in Mercury's evidence chain. The agent answers, books appointments, and captures deposits. Mercury turns each call into a policy-bound, evidence-backed, defensibly billable work event.

312
calls answered
47
appointments booked
12
deposits captured
8
no-shows recovered

Month-end: Northwind bills Beacon and can show exactly what was billable and why. When an audit opens, they export a signed evidence package in minutes, not weeks.

Voice AI Platforms
Vapi · Retell · Bland · LiveKit · Synthflow
Healthcare Systems
Epic (FHIR R4) · AthenaHealth · Nextech · Weave
CRM & Business
Salesforce · HubSpot · Zendesk
Payments & Comms
Stripe · AWS SES · Twilio · SendGrid
Connectivity
HTTPS Webhooks · FHIR R4 · OAuth SMART · Custom REST
Live Enforcement

Real-time policy enforcement in action

Agent Work Receipt evaluates every agent interaction against your compliance policies and rule packs. Below is a real-time conversation between a voice agent and a patient, demonstrating how speech-to-text requests for patient records are intercepted, sensitive data is redacted, and every authorized action is securely logged.

Real-time Policy Engine
Enforcement Active
AI Agent
Idle
Inbound Request JSON
Waiting for agent interaction...
Compliance Boundary Cleartext
Patient Name John Doe
SSN XXX-XX-1234
Medication Metformin 500mg BID
Diagnosis Type 2 Diabetes
Authorized & Logged
HIPAA
SOC 2
Markets

Built for the work AI agents actually do

Agent Work Receipt starts where trust is not optional: regulated and high-accountability agent workflows. Healthcare, insurance, financial risk, payroll, legal, and customer operations are the wedge. Every AI agent builder, SaaS platform, and usage-billed agent product that needs runtime trust to ship into enterprise buyers is the platform opportunity.

Healthcare AI
$750M–$2B TAM

Voice agents, intake automation, revenue cycle, prior authorization, insurance verification, and patient communications. HIPAA requires per-action evidence. 42 CFR Part 2 has been actively enforced since February 2026. Agent Work Receipt is how healthcare AI ships into clinical environments without a compliance engineering team.

HIPAA 42 CFR Part 2 Prior Auth Revenue Cycle PHI Evidence
Insurance & Claims
$400M–$1B TAM

Claims automation, underwriting assistance, brokerage workflows, and payer/provider call operations. Agent decisions need defensible audit trails, decision provenance, and reviewer approval records for regulators and customers who dispute outcomes.

Insurance Reg Claims Evidence Decision Trails Exception Queues
Financial Risk / AML
$300M–$900M TAM

KYC, KYB, AML screening, sanctions review, and financial crime operations. Every agent-assisted review event needs regulatory-grade evidence: reviewer controls, policy versioning, and exportable audit trails that survive regulatory examination.

GLBA AML Evidence KYC / KYB Audit Exports
Payroll / EOR / Legal
$300M–$900M TAM

Worker classification, EOR workflows, tax filings, regulatory submissions, and legal/compliance review. Agents making consequential decisions need decision provenance, human review states, jurisdictional retention, and customer-facing evidence summaries.

Worker Classification Tax Compliance Regulatory Filings Approval Records
Customer Operations
$750M–$1.7B TAM

Payments, refunds, escalations, commitments, service-level guarantees, and billing disputes across contact centers, BPOs, and high-volume customer operations. When customers dispute what your agent did, Agent Work Receipt has the proof. Every billable outcome backed by evidence.

FDCPA TCPA Dispute Evidence Billing Defense
AI Agent Builders & Platforms
$18B–$45B Platform

Every AI agent builder shipping into healthcare, finance, insurance, legal, ecommerce, or high-accountability operations needs runtime trust as a feature. Agent Work Receipt embeds in your stack so you can sell into regulated buyers without building compliance, evidence, retention, and billing-grade event export from scratch. Ship runtime trust as a feature, not a roadmap item.

Vapi Retell ElevenLabs Bland Synthflow OEM / Embed
The runtime trust market opportunity (2030 estimates)
Beachhead
$2.5B–$7.5B
Platform
$18B–$45B
Expanded platform
$40B–$95B+
Long-range
$75B–$150B+
Integration

One trust layer, every architecture

Agent Work Receipt supports internal agent fleets, external AI platforms, and embedded multi-tenant deployments. The same policy engine, evidence vault, and billing-grade event feed scales from a single agent to thousands of tenants across any platform.

mTLS Internal Auth Direct Integration

Internal Agent Fleet

Healthcare AI companies & internal operations teams

Your organization builds its own agents that book, collect, update records, and trigger revenue events. Agent Work Receipt provides the mTLS certificate tree so every internal interaction is captured, policy-checked, and retained as evidence.

If your AI agents book, collect, update records, resolve tickets, or trigger invoices, Agent Work Receipt gives you the evidence trail to prove the work happened, satisfy compliance review, bill for it, and get paid without disputes.

AI Agent
AWR
mTLS · Policy Engine
AthenaHealth
Weave
Salesforce
Webhook Verification OAuth SMART Cross-Platform

Voice AI Agencies

Agencies building on Vapi, Retell, ElevenLabs & Bland

You build AI agents on third-party platforms for clients in healthcare, finance, and operations. Agent Work Receipt acts as the trust broker, enforcing client-specific policy rule packs per tenant and giving you a signed evidence record for every billable outcome you deliver.

Serve regulated clients across verticals without building compliance internally. Add Agent Work Receipt to your stack and unlock every regulated enterprise deal your competitors cannot touch.

Vapi
Retell
ElevenLabs
AWR
Webhooks · FHIR R4
Epic EHR
Salesforce
Stripe
Tenant Isolation API Key Scoping CaaS

Platform Builders

AI-native SaaS platforms & multi-tenant agent products

You are building a platform where your customers deploy their own agents. Agent Work Receipt provides tenant-isolated policy enforcement, per-tenant audit ledgers, and API-key-scoped evidence vaults so your platform ships runtime trust as a feature instead of a roadmap item.

Ship runtime trust as a feature, not a roadmap item. Agent Work Receipt embeds in your stack and produces the evidence your customers will demand when they reach regulated buyers.

Tenant A
Tenant B
Tenant N
AWR
Per-tenant isolation
HighLevel
Zendesk
HubSpot
For payments and settlement infrastructure

Your agents are initiating payments. "The agent said so" is not enough.

When an AI agent initiates a stablecoin transfer, triggers an invoice, or executes a delegated purchase, every counterparty in that workflow needs to answer the same questions before money moves: Was the work real? Was the agent authorized? Did KYC, KYT, and sanctions checks pass? Was policy followed?

Payment rails execute. Agent Work Receipt governs. It is the evidence layer that sits above custody, KYC, and rail infrastructure and produces the signed proof record that justifies each payment.

What rails answer
How do I move USDC, PYUSD, RLUSD, ACH, wire, or card?
What AWR answers
Was this work real, authorized, policy-compliant, and defensible before any payment executes?
The Settlement Receipt primitive
One signed artifact. Work and payment, linked.

The Settlement Receipt links a completed, policy-compliant agent action to a specific payment event: the invoice, rail reference, stablecoin transaction, or ACH trace. It is the artifact that survives audit, GENIUS Act compliance review, dispute, and regulator inquiry.

What a Settlement Receipt contains
Agent Work Receipt reference (the underlying proof of work)
Authority record (delegated mandate, user consent, policy version)
Payment linkage (invoice ID, rail reference, stablecoin tx hash)
Compliance decision (policy checks passed, counterparty checks, KYT state)
Dispute state and credit/adjustment history
Agent Work Receipt decides whether the payment should occur. Bridge, Rain, Fireblocks, ACH, card, or wire executes it. Agent Work Receipt does not move money or custody assets.
For A2A, MCP, and AP2 builders

Your protocol moves tasks between agents. Who is accountable when something goes wrong?

A2A, MCP, and AP2 protocols solve task routing and tool invocation between agents and systems. They do not solve the evidence problem: when an agent receives a mandate from another agent, invokes a tool, produces an artifact, or hands off a task, there is no neutral proof object that records authority, what was requested, what was produced, which policy applied, and whether the result is billable or disputable.

Protocol Receipts attach to the underlying Agent Work Receipt and record the cross-agent evidence that protocols omit.

The problem protocols leave unsolved
Which agent originated the mandate, and did it have authority?
Which artifact was produced, and is it the artifact that was requested?
Which policy version ran across the full task chain?
If the task is billed, which agent or platform receives the credit?
If something fails or is disputed, where does accountability land?
What a Protocol Receipt records
Mandate: originating agent, delegated authority scope, user consent record
Participants: all agents and platforms in the task chain
Artifact lineage: what was produced at each step and by whom
Task state: open, completed, handed off, failed, disputed
Policy decisions across every agent in the chain
Protocol Receipts attach to Agent Work Receipts. One canonical proof object covers the full agent chain: who did what, on whose authority, under which policy, and with what result.
Regulatory coverage

One policy engine, configurable by industry

Agent Work Receipt's policy engine attaches vertical rule packs to every agent action at runtime. Each pack governs which evidence is required, which disclosures must occur, which data must be redacted, and what retention applies. Start with healthcare, add packs as you expand to new verticals.

HIPAA / HITECH

Healthcare privacy, substance use disorder protections, FTC Health Breach, FDA off-label communication rules.

FDCPA / Reg F

Debt collection communication controls. Call frequency, disclosure requirements, mini-Miranda, time-of-day restrictions.

GLBA

Financial data protection and privacy. Safeguards Rule, privacy notice requirements, information sharing controls.

PCI DSS

Payment card data handling. Scope reduction, access controls, audit trails for payment-taking agent interactions.

TCPA

Telemarketing consent and calling rules. Prior express consent, autodialer restrictions, DNC list management.

Insurance

Claims communication and disclosure requirements. State-specific insurance regulations, prohibited practices, documentation rules.

42 CFR Part 2

Substance use disorder record confidentiality. Active OCR enforcement since February 2026. Stricter disclosure rules than standard HIPAA.

AI Governance

EU AI Act, NIST AI RMF, ISO 42001, Colorado AI Act. Transparency, explainability, and documentation requirements for regulated AI deployments.

Plans

Pricing

Four editions: Ledger, Guardian, Citadel, and Enterprise. Priced by edition depth and policy depth. Monthly commitment, included allowances, and predictable overages.

See Full Agent Work Receipt Pricing

Hello Voice Performance-based includes AWR Core at no incremental charge. Hello Voice Enterprise includes a contracted Agent Work Receipt allocation sized to scope. View Hello Voice pricing

Certification

Agent Work Receipt Compliance Certification

Certify your platform, agents, and deployments through continuous compliance scanning.

Level 1
AWR Certified

Foundation scanning passed for 90+ consecutive days with no critical violations. Foundation certification for platforms beginning compliance adoption.

Level 2 / Recommended
AWR Certified, Advanced

Enforcement scanning for 90+ days with compliance score above 95%. Full policy enforcement, audit evidence generation, elevated governance requirements.

Level 3
AWR Certified, Premier

Full platform scanning for 180+ days with compliance score above 98%. Complete audit support and priority certification review.

Contact sales for certification pricing and eligibility.

FAQ

Agent Work Receipt. Runtime trust infrastructure for AI agents. Common questions from compliance, operations, finance, and platform teams.

Agent Work Receipt is the durable unit of truth for economically meaningful agent work: the compliance and revenue evidence layer that proves AI-agent work was completed, policy was followed, and billing is defensible. Agent Work Receipt receives interaction data from any AI platform via webhooks or mTLS, records events in an immutable ledger, enforces configurable policy rule packs (HIPAA, FDCPA, GLBA, PCI DSS, TCPA, 42 CFR Part 2, Insurance, AI Governance), validates whether work events are evidence-backed and billable, and produces audit-ready evidence packages for compliance review, dispute resolution, and customer transparency. Agent Work Receipt produces the proof of work record. Your billing system prices it.
No. Agent Work Receipt is platform-agnostic and connects to any AI agent platform via HTTPS webhooks. Hello Voice Performance-based customers get AWR Core embedded at no incremental charge: event ledger, encrypted retention, HIPAA baseline scanning, and immutable audit logging. Hello Voice Enterprise includes a contracted Agent Work Receipt allocation sized to scope. Agent Work Receipt is also available standalone for any builder, agency, or platform.
Agent Work Receipt supports HIPAA/HITECH (healthcare), FDCPA/Reg F (debt collection), GLBA (financial services), PCI DSS (payments), TCPA (telemarketing), Insurance, 42 CFR Part 2 (substance use disorder), and AI Governance (EU AI Act, NIST AI RMF, ISO 42001, Colorado AI Act) rule packs. New verticals are added as configurable rule packs within the same policy engine.
Agent Work Receipt standalone is priced by edition: a monthly platform commitment (AWR Ledger from $499/mo, AWR Guardian from $1,500/mo, AWR Citadel from $7,500/mo) plus included monthly allowances of trusted interactions, evidence storage, and policy evaluations. Overage is billed at flat unit rates. AWR Core is embedded in Hello Voice Performance-based at no incremental charge. All editions are implementation-based: onboarding completes before billing begins.
Agent Work Receipt gives compliance teams an audit-ready record of what happened, which policy applied, who accessed the evidence, and why each event was permitted or escalated. Policy is attached to agent work at runtime, not reconstructed after the fact. When an audit opens, Agent Work Receipt assembles a signed evidence package per event: the interaction record, the policy version that applied, the decision verdict, and the evidence retained under the correct regulatory retention rule. Audit answer time goes from weeks to minutes.
Agent Work Receipt gives revenue teams a defensible work record behind every AI-agent charge: what happened, what evidence supports it, whether the system-of-record write confirmed, whether the event was duplicate-safe, and why it can be billed. "Billable event" and "defensibly billable event" are not the same statement. Agent Work Receipt closes the gap so your billing runs on events that can survive a dispute, a payer audit, or a customer inquiry without restatement.
You can build an event log. Building an evidence layer is different. Runtime trust requires: a policy engine that evaluates regulatory rule packs at sub-50ms, per-event decision receipts with cryptographic identity, multi-framework retention rules that satisfy healthcare, finance, and insurance simultaneously, a billing-grade event feed with duplicate suppression and system-of-record write confirmation, and audit export in minutes, not weeks. That is 12+ months of compliance engineering. Agent Work Receipt embeds in your stack in under a week and ships that layer as a feature. Your enterprise buyers will demand it. Your competitors will be waiting on procurement questionnaires you cannot answer.
Payment rails answer "how do I move money." Agent Work Receipt answers "why should this payment occur." When an AI agent initiates a stablecoin transfer via Bridge, Rain, Fireblocks, or any other rail, Agent Work Receipt provides the evidence layer above the rail: was the work real, was policy followed, did the delegated agent have authority, and is the billing defensible? The Settlement Receipt primitive links verified work to a specific payment event in a signed artifact that survives audit, dispute, and regulatory review under the GENIUS Act framework. Agent Work Receipt is rail-agnostic: USDC, PYUSD, RLUSD, ACH, wire, and card are all supported on the payment side.
Mercury Nexus is the compliance-governed multi-agent orchestration exchange: a neutral layer where AI agents on different platforms hand off work with every action governed, logged, and audit-ready. The VISA analogy applies: Nexus becomes the trust layer between agent platforms the way payment networks became the trust layer between issuers and merchants. Four primitives make it work: Task Lifecycle Registry, Agent Registry, Event Bus, and Handoff Protocol. Nexus is the V5 to V6 roadmap milestone. V4 (real-time compliance gateway, in progress) is the prerequisite. Nexus beta is available to qualified platform partners; general availability follows. Contact sales for roadmap briefing and early-access qualification.

Prove what your AI agents did. Enforce policy. Defend billing.

Agent Work Receipt turns every agent action into evidence-backed proof: the work happened, policy was followed, billing is defensible. One integration. Every platform. Connect your first agent in under a week.