Synthetic documents for safer testing.

Test AI systems, workflows, and software with structurally consistent synthetic documents and deliberate edge cases, without using production data.

Every package includes human input, review, and creative scenario design, ensuring the documents are never generated by AI alone.

Try for free Browse datasets

Finance & Invoices

invoice_batch_Q4_2024.zip

pre-generated dataset pack

Realistic structures with deliberate edge cases for dependable testing.

Try Deniable for free

Get a free synthetic sample.

Choose a testing category and inspect a randomly selected pre-generated document before you buy. You can see the real format, metadata, and edge-case structure in a controlled account-gated sample pool.

Use the web sample for a quick look, or try the authenticated API sample path with no credits charged.

TRY A FREE SAMPLE

What Deniable provides

Test inputs built for the cases routine data misses.

Document cases

Pre-generated inputs for difficult testing jobs.

Human involvement

Every package includes human input and scenario design.

Two delivery paths

Choose a fixed dataset pack or prepaid API credits.

Repeatable access

Filter, retrieve, and test the cases your workflow needs.

Production data carries risk. Empty test sets leave blind spots.

Production documents contain real names, account numbers, and legal obligations. Deniable provides structurally consistent synthetic documents with deliberate edge cases for testing AI systems, data workflows, and software safely.

Documents for the cases your systems need to handle.

Choose a category that matches your workflow, then test against realistic structures and intentionally difficult cases.

Dataset category

Support Tickets

Dataset category

Product Reviews

Dataset category

Social & Chat Logs

Dataset category

Meetings & Call Transcripts

Dataset category

Legal Documents

Dataset category

HR Documents & Resumes

Dataset category

Finance & Invoices

Dataset category

Code Q&A

Dataset category

Long-form Documents & Ebooks

Dataset category

Mixed Edge Cases

Dataset category

Specialty Collections

The Deniable approach

Useful structure, difficult conditions, clear access.

Deniable gives teams a repeatable way to test the systems that have to interpret, route, retrieve, or transform documents.

Pre-generated documents

Start with ready-made synthetic documents instead of exposing production records or preparing every case by hand.

Human scenario design

Human input and creative review shape the situations around each document, so the cases are not AI output alone.

Deliberate edge cases

Missing context, conflicting details, and awkward structures make difficult test conditions visible before they reach production.

ZIP or API access

Use a fixed dataset pack for a complete download, or prepaid API credits when your workflow needs retrieval and usage tracking.

How it works

Choose a test pack

Filter by category, format, and language. Every dataset is pre-generated and immediately available for download.

Use the delivery method that fits

Download a ZIP directly from the store, or connect via the REST API or MCP server for CI/CD integration.

Test where failures matter

Use the documents in AI systems, data workflows, product QA, or any process that benefits from realistic, difficult inputs.

Questions before you start

Clear answers about access and use.

Do I need an account before I can buy a dataset?

Yes. An account and verified email are required before checkout or API access. Sign in or create an account to continue.

What is the difference between a dataset pack and API credits?

A dataset pack is a fixed product delivered as a ZIP. API credits are prepaid and let an authenticated client retrieve documents from the curated catalogue through the API. They are separate products with separate access flows.

Are the documents generated only by AI?

No. Every package includes human input, review, and creative scenario design. The human contribution is part of the product principle, not an afterthought to raw generation.

Can the datasets be used for model training or fine-tuning?

The intended internal license permits development, QA, evaluation, commercial internal use, model training, and fine-tuning. Redistribution, resale, public hosting, and sublicensing remain outside that internal license.

Do repeated API retrievals use credits again?

Yes. Each document delivery consumes credits, including a repeated retrieval of the same document. This keeps usage accounting simple and helps prevent uncontrolled reuse.

What happens when an API account has no credits left?

The API does not deliver another document when the balance is insufficient. Add a new prepaid credit pack to continue; balances do not go negative. Automatic renewal is disabled, so credits are never replenished or charged automatically.

Dataset packs

Get the complete case set.

Fixed dataset packs for teams that want a ready-to-download collection.

Browse datasets

API credits

Retrieve cases in your workflow.

Retrieve pre-generated cases as JSON with prepaid credits. No monthly subscription, and no generation pipeline to maintain.

BUY API CREDITS