# 48-Hour Paid Proof Pilot

> Record one workflow on a test copy of your system. In 48 hours we compile the program, bind a second read of the target record, test planted faults (wrong values, blanked fields, missing saves, duplicates, sibling edits and wrong targets) and deliver signed record/run/v1 receipts before anything touches production.

- Canonical: https://openadapt.ai/pilot
- Updated: 2026-11-22

## What happens in 48 hours

| Window | Work done | Output |
| --- | --- | --- |
| Day 1 | Record one real workflow on your test copy and bind a read-only second read of the target record. | Compiled workflow program and verification contract qualified on your test copy. |
| Day 2 | Run planted-fault checks (wrong values, blanked fields, missing saves, duplicates, sibling edits and wrong targets) alongside clean-write runs. | Cryptographically signed `record/run/v1` [receipts](https://openadapt.ai/receipts) and a fault-catch summary you can verify offline. |

## What you keep

- A compiled program for your one workflow that checks the target record before saving and reads the saved record back afterward.
- A verification contract bound to your read-only second read path, with no app-specific engine forks.
- A planted-fault audit summary showing every injected fault class tested and whether each fault was stopped.
- Cryptographically signed `record/run/v1` [receipts](https://openadapt.ai/receipts) for both verified and stopped runs, verifiable offline before deciding on production rollout.

## Deployment and data boundaries

| Option | How it runs | Data boundary |
| --- | --- | --- |
| Your VPC or workstation (BYOC) | OpenAdapt and its customer-run connector run inside your own cloud account or on your machine against your test copy. | Credentials and records stay in your environment. |
| Local-model zero-egress | Runs locally with the local-model runtime and offline receipt verification. | Zero external model calls and zero outbound record egress. |

See [Security](https://openadapt.ai/security) for how credentials, read-only verification queries and offline receipts are separated.

## Scope and commercial terms

- One workflow on a test copy or staging environment with synthetic or test records.
- A read-only second read path (database query, API read-back or structured export) confirmed on the intake call.
- Fixed-fee paid engagement scoped with you on the 30-minute intake call before work begins, and credited toward a Team or Enterprise deployment when you proceed.
- If your system holds patient data or falls under healthcare regulation, the pilot runs in your environment and feeds directly into a full [qualification sprint](https://openadapt.ai/services).

## Start your 48-hour pilot

Tell us the workflow and the system it writes to on [the booking page](https://openadapt.ai/book). On a 30-minute call with OpenAdapt's founder, we confirm the second read path, agree on the pass criteria and schedule Day 1.

[Book a 30-minute call](https://openadapt.ai/book) · [See pricing](https://openadapt.ai/pricing) · [How a qualification sprint goes](https://openadapt.ai/services)
