Choose your starting point
Define a new document or scenario from scratch, or use a GritRedact output as a template to preserve a familiar layout and structure.
GRITRENDER / SYNTHETIC EVALUATION DATA
Generate synthetic documents, images and audio with known answers.
Create synthetic documents, images, faces, fingerprints and audio for development, training and testing. Control the fields, business rules and variations your workflow needs—even when the right production samples don’t exist.
SCENARIO REQUIREMENTS
↓ GENERATE
SYNTHETIC CONTENT + KNOWN VALUES
The workflow
Design the data. Define the rules. Generate the scenarios.
Start with a blank template or a reviewed, sanitized seed. Use the web designer to configure the content, then generate the volume and variation needed for your evaluation.
Set fields and data rules
Define fields such as names, dates, account numbers and amounts. Set lengths, ranges and relationships—for example, an age rule or a ten-digit account number.
Control volume and variation
Generate positive and negative test scenarios, from clean samples to damaged scans, handwritten fills and adversarial patterns.
What you can generate
Documents and forms
Statements, claims, KYC documents and multi-page forms with realistic layouts and consistent values across related files.
Identity records and images
Synthetic IDs, certificates and image artifacts for onboarding and verification tests, without using real individuals’ records.
Faces and fingerprints
Vary synthetic faces and fingerprints for identity-verification tests, including the same synthetic identity at different ages.
Text and audio
Customer-service transcripts and synthetic voice interactions across intents, accents and difficult audio conditions.
Output formats
PDFJPEGPNGTIFFBMPGIFWebPMP3
Build the cases your test set is missing
30+ negative document variants
Test OCR and extraction with handwriting, ink stains, torn edges, fax artifacts, photocopies and smartphone-photo conditions.
Consistent data and known results
Set field values and business rules, keep related files consistent, and compare system output with known inputs.
Business-rule and specification inputs
Build positive, negative and adversarial cases from business rules, schemas or API contracts—even before production data exists.
Invisible watermarking
Verify an asset’s GritRender origin with an invisible watermark embedded at generation, including when shared with other teams.
Illustrative samples of the same fictional record under different visual conditions.
Where it fits
Test the conditions your system will encounter.
Evaluate OCR, document extraction, speech and conversational systems against controlled inputs. Create rare cases and difficult conditions that are expensive to assemble manually or absent from production samples.
Send generated PDFs and images through GritTransform or your existing extraction system, then compare the returned fields with the known values. Use GritRedact to prepare approved real content before using it as a seed.
- Handwritten, scanned and degraded document inputsOCR / IDP
- Known-answer scenarios for training and evaluationAI / Agents
- Transcripts, intents and voice interactionsConversation
- Fraud patterns, outliers and negative casesRobustness
- Synthetic datasets for vendor and distributed-team testingSharing
Available on-premises and as SaaS
Run GritRender offline on your infrastructure, or use a sales-assisted SaaS subscription. Design content through the web platform and connect generation to your workflows through APIs.
We’ll help define templates, rules, output formats and test coverage, then choose the deployment that fits your team. Enterprise options include private cloud, VPC and air-gapped environments.
Before you get started
Do we need production data to get started?
No. Start from a blank template, field schema or business scenario. A sanitized seed is an option when you want to retain a real document’s layout and structure.
Can we use Redact and Render together?
Yes. Review and sanitize approved source content with GritRedact, then use that output as a seed for synthetic expansion. Configure additional samples, variations and negative cases in GritRender.
How do we know what the test result should be?
Define the values and rules used to generate the content. They provide the expected results against which you can evaluate extraction, transcription or downstream business behavior.
How should we assess realism?
Evaluate representative outputs against your target workflow. Review document structure, media characteristics and the behavior of your downstream system before expanding the test corpus.
What should we bring to a demo?
Bring a document type or scenario, required fields, business rules, expected volume and the failure cases you want to test. We’ll help translate those requirements into a generation workflow.
Explore the evidence
Inspect examples before defining your test set.
Start with the existing fictional document examples and known values, then review the relevant enterprise workflow.
OCR test examples with known values
Inspect low-light, blur, fax and photocopy conditions with the reference JSON.
Laboratory requisition OCR test coverage
See the existing evaluation of controlled document variants and expected results.
Consistent loan-document test files
See the existing redact-then-generate workflow for financial-services testing.
START WITH DOCUMENTS YOU CAN EXPLORE
Bring a sample into your workflow.
Free PDF packs for evals, sandboxes and testing: readable and degraded documents with reference fields. Need implementation or operational help? Discuss support around GritWorks products.
Start with your workflow
Put your unstructured data to work.
Tell us what your team needs to achieve. We’ll help you choose the right products, work through your requirements and explore on-premises or SaaS deployment.
Request a working session