# AugmentEV — confidential, provable artificial intelligence for sensitive data (Paseo)

> AugmentEV runs artificial intelligence on data you can't expose, and produces a post-quantum-signed, tamper-evident Proof-of-Task-Execution (ML-DSA-65) for supported completed jobs. Standard provides general-purpose execution with a signed receipt. Confidential adds a sealed, hardware-attested Intel TDX boundary designed to isolate inputs and outputs from the host and operator during execution. The public verifier is operated by AugmentEV and checks receipt signatures without requiring an account. Roadmap capabilities are not represented as live. Canonical machine contract: https://augmentev.com/.well-known/agent.json

## Start here
- [Verify a receipt](https://augmentev.com/verify.html): load a live receipt → valid: true; tamper with one character → valid: false. Public, no account.
- [Signed agent catalog](https://augmentev.com/.well-known/agent.json): machine-readable list of agents, endpoints, auth, and pricing.
- [Agent reference](https://augmentev.com/paseo/docs/agents.html): Khanoor (SOC), Resolve (support), DocFlow (documents) — capabilities, endpoints, and comparison.
- [Quickstart](https://augmentev.com/paseo/quickstart.html): from signup to first paid call.
- [Catalog](https://augmentev.com/paseo/catalog.html): browse available agents.
- [Pricing & signup](https://augmentev.com/pricing.html): per-call tiers and self-serve key minting.
- [Services](https://augmentev.com/services.html): AI-agent evidence readiness — assessment, implementation, retainer.

## The evidence rail
- Supported completed jobs return a tamper-evident, post-quantum-signed receipt (ML-DSA-65); receipt records are retained in tamper-evident form for at least six months.
- Public verification endpoint: POST https://api.augmentev.ai/v1/evidence/verify — an AugmentEV-operated endpoint that checks a record without requiring an account.
- Live today: hardware-attested confidential execution (Intel TDX), signed receipts for supported completed jobs, public verify.
- Design of the full rail (not yet live): countersignature by an independent custodian's hardware security module, and write-once neutral custody — records no party to a job can alter, reorder, or drop.
- For agent operators: prove your agent's conduct to auditors, customers, and counterparties.
- For auditors and customers: check receipt signatures through the AugmentEV-operated public verifier; this is not independent custody.

## Services — AI-agent evidence readiness
For AI-agent operators, insurers, and regulated enterprises who must prove conduct to underwriters, auditors, or claims adjusters.

### What we deliver — mapped to AIUC-1
AIUC-1 organizes agent risk into six pillars: Data & Privacy · Security · Safety · Reliability · Accountability · Society. AugmentEV prepares the technical evidence backbone:
- Evidence inventory & gap assessment against all six pillars
- Tamper-evident activity logging with post-quantum-signed (ML-DSA-65) receipts
- Incident & failure-response documentation
- Evidence packs assembled for underwriter questionnaires, quarterly re-tests, or claims

### Engagement tiers
| Tier | Fee | Duration | You get |
|------|-----|----------|---------|
| Evidence Readiness Assessment | $5,000 fixed (founding; $10,000 standard) | 2 weeks | Gap report, evidence inventory, prioritized remediation plan |
| Implementation Sprint | $15,000–$30,000 fixed (founding; $25–50K standard) | 4–8 weeks | Live tamper-evident pipeline, independent custody design, first evidence pack |
| Evidence Retainer | $3,000 / month (founding; $6,000 standard) | Ongoing | Quarterly refresh, renewal support, claim-time packs |

**Independence note:** AIUC-1 is created and administered by the Artificial Intelligence Underwriting Company (AIUC); certificates are issued only by AIUC through its accredited auditors. AugmentEV is independent and unaffiliated. We prepare your evidence; we do not certify or guarantee certification.

Contact: bassem@augmentev.com · https://augmentev.com/services.html

## The Paseo platform (self-serve)
- Self-serve access to AugmentEV's own verifiable agents: sign up in one request, call over REST or MCP, pay per call — supported completed jobs return a post-quantum-signed proof. Third-party publishing is on the roadmap, not yet open.

## Interfaces
- REST: https://api.augmentev.ai/api/v1
- MCP: https://api.augmentev.ai/mcp
- Signup (one request, returns a Bearer API key): https://api.augmentev.ai/v1/signup
- Machine-readable price card: https://augmentev.com/.well-known/pricing.json
- Verify: https://api.augmentev.ai/v1/evidence/verify

## Available agents
- **khanoor** — Security Operations (SOC): multi-signal telemetry ingestion, corroborated threat triage with MITRE ATT&CK mapping, and containment-capable response (webhook/noop by default; provider-configured per deployment).
- **resolve** — Customer Support: intent classification, retrieval-grounded answers, and smart escalation with full context packages.
- **docflow** — Document Processing: OCR in 163+ languages, hybrid entity extraction, classification, and cryptographically-signed audit receipts.

## How it works
1. Discover — read the signed catalog at /.well-known/agent.json. No sales call.
2. Sign up — one request mints an API key. No contract, no seats. Current founding discount: 50% off self-serve rates for the first three months.
3. Call — invoke an agent over REST or MCP; set per-call cost and runtime caps.
4. Pay per call — usage-metered billing.
5. Verify — supported completed jobs return a post-quantum-signed proof of execution (ML-DSA-65, FIPS 204).

## Tiers
- Standard — general-purpose verifiable agent execution; supported completed jobs return a signed proof; usage-based billing.
- Confidential — hardware-attested confidential execution (Intel TDX), fail-closed, with a post-quantum-signed proof. FIPS 140-3 validation is on our roadmap. For regulated, audit-first teams in health, legal, and financial services.

## Provider
- AugmentEV, Inc. — https://augmentev.com
- Support: support@augmentev.com
- Security: https://augmentev.com/.well-known/security.txt