Launch film
Why we’re building Ample.
Receipts for enterprise AI
A receipt for every
AI answer.
Tamper-evident receipts for AI answers.
You keep the record. Your auditor can check it.

“the model said so”
is not an answer.
AI you can show
an auditor.
AI drafts credit memos, answers customers and flags transactions. Later, someone needs to check what happened.
Editable logs and changing model versions leave gaps. Ample records the answer and its context as it happens, so an auditor can check it independently.
Explore the record: Capture, Inspect, Replay and Chain
The request, answer and context are recorded together.
Inside your environment.
Your application sends requests through Ample to your models. Answers return through Ample to your application. Ample records the answer and its context in your storage. Select a component to read its role.
Your infrastructure Illustrative on-prem deployment
Select a component for details.
Answer receipts
Keep receipts in your own storage. Your auditor can verify and replay independently.
Add Ample directly, or through your sovereign or on-prem AI provider. Receipts stay in your storage.
See what is recorded
- Input
- Request hash; original input if policy allows
- Output
- Answer hash and reasoning trace
- Model
- Model, version, configuration and weights/checkpoint identifier
- Time
- When the answer was produced
- Policy
- Rules and settings applied
- Chain link
- Previous receipt’s hash
A separate verifier lets your team, auditor or supervisor check the chain and replay recorded decisions.
Before you deploy
A clearer record.
A few clear answers.
What Ample captures, where it runs, and what your auditor can check.
Explore the deploymentWhat is Ample?
Ample, built by ISRM Labs AB, records AI requests, answers and their context as tamper-evident receipts. Your team keeps the records in its own storage so an auditor can check what happened independently.
What does an AI receipt contain?
An Ample receipt records the request hash, answer hash and reasoning trace, model version and configuration, timestamp, applied policy, and the previous receipt’s hash. Original inputs are retained when your policy allows.
How do tamper-evident receipts differ from editable logs?
Each receipt includes the previous receipt’s hash, linking the records into a chain. Changing or removing a record leaves a break that a separate verifier can check. This gives an auditor a way to check the record independently of an editable log.
Where can Ample run, and where do receipts stay?
Ample can run on your own servers, in your private cloud, or alongside inference on your own GPUs. It works with hosted, open-weight and on-prem models. Receipts stay in your storage. You can work directly with Ample or through your sovereign or on-prem AI provider.
What can an auditor check?
Your team, auditor or supervisor can use a separate verifier to check the chain and replay recorded decisions. The receipt provides the model, inputs, policy, output and timestamp associated with the answer.
Who is Ample designed for?
Ample is designed for teams that need to stand behind enterprise AI decisions, including banks and lenders, funds and asset managers, insurers, and public-sector organisations. Examples include model-risk review of credit and AML assistants, research and reporting, claims and underwriting, and case-handling AI.
Does an AI receipt prove that an answer is correct?
A receipt documents the answer and its recorded context so they can be checked. It does not by itself establish that the answer is factually correct or that a deployment meets a particular regulation.

From black boxesto open ledgers.
Bring Ample to the AI your organisation needs to stand behind.
