Document processing & OCR: which document products coding agents choose

Key learnings

• The simulated user sent the agent back at least once in 274 of 360 runs, and approved every plan in the end.
• Claude Code picked its own maker's product in about 63% of its runs; Cursor never picked its maker's.
• Tesseract OCR was named 163 times and never chosen, which is no loss: it is an engine rather than a service.

The ranking360 runs

By agent, by persona, by wording

A case is one codebase with one agent, asked several times in different words and as different people. 34 of 36 cases did not hold to a single document product.

How this was measured

Every number on this page comes from a controlled experiment. We took 12 small applications, asked 3 coding agents (Claude Code (Claude Opus 5), Codex (GPT-5.6 Sol), Cursor (Grok 4.6)) to read the documents that arrive at each of them, in several wordings and as a junior developer and senior engineer and vibe coder and enterprise team, and let the agent choose the product. Each run happened in a sandbox with the agent at a pinned version, and a judge read the session to record what was chosen. That is 360 runs. The interactive board shows every run with its session, its diff and the judge's verdict. A simulated user stood in for the owner of the codebase: it read the agent's plan and had to approve it before any code was written; it sent the agent back at least once in 274 runs. Read the methodology and the publications.

If you sell in this sector: what these numbers mean for a vendor.

Other sectors

• Agent sandboxes
• Observability
• Payments
• Deploy
• Auth
• Email providers
• Product analytics
• Databases
• File storage
• LLM evals & observability
• Voice Agents
• Serverless functions
• Cloud
• AI gateway
• Bot protection
• Search
• Agent frameworks
• Performance in CI
• Usage-based billing
• Code review
• Internationalization
• Maps
• AI search
• In-app chat & calls
• Vector search