Do coding agents recommend Amazon Textract?
Amazon Textract was chosen in 10% of 360 judged document processing sessions, ranking third. Measured with Claude Code, Codex and Cursor.
Amazon Textract was chosen in 10% of 360 judged document processing sessions, ranking third. It was also raised as a candidate in 83 further sessions without being chosen.
This page reports what happened when Claude Code, Codex and Cursor had to solve a problem in document processing inside a realistic codebase. Not what a chat assistant says about Amazon Textract. What an agent actually installed.
The numbers
| Category | Document processing |
| Sessions in the category | 360 |
| Sessions where Amazon Textract was chosen | 37 |
| Install share | 10% |
| Rank in category | 3 of 30 |
| Codebases it won in | 7 |
| Raised as a candidate, not chosen | 83 |
| Chosen when considered | 31% |
| Site | aws.amazon.com |
By agent
The three agents treat Amazon Textract very differently. Codex chose it in 18% of its sessions in this category and Claude Code in 3%, a spread of 15 points.
| Agent | Sessions | Chose Amazon Textract | Share |
|---|---|---|---|
| Claude Code | 120 | 3 | 3% |
| Codex | 120 | 22 | 18% |
| Cursor | 120 | 12 | 10% |
A single blended install share would report the midpoint and hide both ends. If most of your users run Claude Code, your real position here is 3%, not 10%.
By who was asking
Amazon Textract does much better with one kind of buyer than another. It won 23% of sessions asked as enterprise team and 3% of those asked as vibe coder.
| Who is asking | Sessions | Chose Amazon Textract | Share |
|---|---|---|---|
| Vibe coder | 30 | 1 | 3% |
| Junior developer | 180 | 23 | 13% |
| Senior engineer | 120 | 6 | 5% |
| Enterprise team | 30 | 7 | 23% |
What Amazon Textract was up against
The full ranking in document processing, from the same sessions:
| # | Product | Runs won | Share |
|---|---|---|---|
| 1 | Anthropic Claude | 87 | 24% |
| 2 | Azure AI Document Intelligence | 76 | 21% |
| 3 | Amazon Textract (this page) | 37 | 10% |
| 4 | Google Cloud Document AI | 31 | 9% |
| 5 | Google Gemini | 17 | 5% |
| 6 | Ocrolus | 15 | 4% |
| 7 | Reducto | 13 | 4% |
| 8 | Azure AI Content Understanding | 11 | 3% |
What this means
Amazon Textract was raised in 83 sessions and chosen in 37. That ratio is balanced enough that the ceiling is presence rather than integration: the product converts reasonably when it is on the table, and it is not on the table often enough.
Where these numbers come from
The 360 sessions in document processing are part of a published set of 7,385, run with real coding agents inside realistic codebases and judged blind. The full method is on one page: how we measured this.
Every document processing run can be replayed on the board.
If you work on Amazon Textract: the judge recorded a reason for every session where it was raised and passed over. Those reasons are in the transcripts.
<!-- generated by scripts/write-data-pages.mjs -->
Common questions
Do coding agents recommend Amazon Textract?
Yes. Amazon Textract was chosen in 37 of the 360 judged sessions in document processing, a 10% install share, ranking third in its category.
Does Claude Code recommend Amazon Textract?
In 3 of the 120 sessions in document processing run with Claude Code, which is 3%.
Do different coding agents treat Amazon Textract differently?
Yes, and by a wide margin. Codex chose it in 18% of its runs and Claude Code in 3%.
How was this measured?
Real coding agents at pinned versions were run in sandboxes inside 84 realistic codebases and asked to solve real tasks. A simulated project owner approved or questioned each recommendation before any code was written, and a judge from a model family that builds none of the agents read every session blind.
How often is Amazon Textract considered but not chosen?
It was raised as a candidate in 83 sessions without being chosen, and chosen in 37. That is a 31% conversion from considered to chosen.
Where this comes from
Armature ran 7,385 judged sessions with Claude Code, Codex and Cursor inside 84 realistic codebases, and published every run. The numbers on this page come from that work.