AI-generated test data, governed by you,
delivered into your systems.

Generate data from a chat prompt

Describe what you need in plain English. DataMaker's AI agent writes the schema, generates the data, and pushes it straight to your database or API endpoint, all from a single prompt.

1. User prompt
2. AI understands schema
3. Data generated
4. Exported / seeded

A generator for every field

Choose from over 50 built-in generators: names, emails, addresses, IBANs, credit cards, SSNs, IMEIs, products, or define custom ones with AI, Python, or your own regex.

Build a schema once, reuse it everywhere

Save your data schema as a reusable template. Share templates across your team, nest objects for complex JSON structures, and generate any number of records on demand, via the UI, API, or AI agent.

1. Define template fields
2. Save template
3. Generate N records
4. JSON · CSV · DB · API

Turn real data into templates automatically

Upload a JSON, CSV, YAML, or OpenAPI file, or point DataMaker at a live database, and let AI reverse-engineer a reusable template from your existing schema. It reads the foreign keys too, so related tables come out in an order that actually loads.

1. Upload CSV / JSON / OpenAPI
2. AI analyses column types
3. Template auto-generated
4. Generate N synthetic rows

Read your databases, seed your endpoints

Connect PostgreSQL, Microsoft SQL Server, Oracle, and MySQL to introspect schemas, walk foreign keys, and extract real rows. Push generated records into REST and OData endpoints with your own headers and auth, or seed PostgreSQL directly.

Built-in SAP integration

DataMaker talks to SAP OData directly. Discover the services on a system from the Gateway catalogue, parse $metadata into a schema graph the agent can actually reason over, apply OData filters, and POST generated records back with automatic CSRF handling.

• OData V2 & V4 (A2X)
• Service discovery from the Gateway catalogue
• Schema graph, with drift detection on $metadata
• Auto-CSRF and per-record POST back

Wired into the tools your testers already run

DataMaker connects to Tricentis Tosca (Cloud and on-prem Commander), Jira Cloud, and Xray. The agent reads test cases and their designs, writes test data back into Tosca test sheets, and files or updates the Jira and Xray artefacts that prove a run happened.

• Tosca Cloud and Tosca Commander on-prem
• Read test-case designs, write test data back
• Jira issues, comments and transitions
• Xray tests, test plans and execution import

Automate complex data workflows

Write Python scenarios to orchestrate multi-step data generation: call APIs, run validations, chain templates, and seed multiple systems in one automated run. Stored, versioned, executable from UI or agent.

Move real data between systems, and prove it landed

Beyond generation, DataMaker runs governed migrations: extract from the source, map with an approved plan, mask, load into the target, then reconcile. Every run is checkpointed, so an interrupted job resumes instead of restarting, and a load can be rolled back record by record.

Plug DataMaker into any AI agent

DataMaker ships MCP servers with 90+ tools across data, SAP, Tosca, and Atlassian. Connect them to Claude, Copilot, Cursor, or any MCP-compatible agent so your AI assistant can generate and deliver test data without leaving the editor.

Your data never has to leave your infrastructure

DataMaker deploys inside your own environment: self-hosted in your data center or private cloud, or as a desktop app on a tester's machine. Credentials, generated data, and the systems being filled all stay behind your firewall.

• Self-hosted deployment with Docker
• Desktop app with a fully local mode
• Credentials stored encrypted, used in-network
• No detour through someone else's cloud

AI you can let near your test systems

Templates and scenarios define which fields, value ranges, and systems the agent may touch. Roles decide who may do what, large writes stop for a human, and every governed tool call is recorded in your own database.

• Role-based access down to the single action
• Big writes need a one-shot human approval
• Audit trail stays inside your deployment
• Read-only mode and a kill switch for the agent

Anonymise production data and keep it joinable

A masking policy assigns a strategy per field: replace with a realistic fake, hash, tokenise, redact, or keep the format valid. Masking is deterministic and keyed on the field class, so the same person masks to the same value in every table and your data still joins.

• 29 PII classes, from IBAN and VAT to VIN and IMEI
• Consistent across tables, so joins survive
• Reversible tokens, behind a separate permission
• Run report as your compliance artefact

Bring your own LLM, including a local one

DataMaker is not tied to one AI vendor. Point it at OpenAI, Anthropic, Azure, Bedrock, Google, or Mistral, at your own OpenAI-compatible gateway, or at a model running on your own hardware through Ollama, LM Studio, or llama.cpp.

• 20 providers, plus any OpenAI-compatible endpoint
• Local models for air-gapped environments
• Keys held server side, never in the browser
• Model choice per deployment, not per vendor contract