Find AI builders by proof, not talk.
Search by what they actually build — tokens, builds, hours, tools — captured live from real coding sessions.
Building with AI yourself? Get found →
0 builders · 19 projects · 6 open to opportunities
0 builders, 19 projects, 6 open to opportunities.
@muzzythinks· 8hDebugged Sentry
Sentry
MARATHON
Most lines changed this week
12,036 lines
13.4Mtokens
39prompts
4:21:45time
12,036lines
Claude Codeclaude-opus-5Markdown$323.70
mostly thinking
4h 21m session running commands and testing. 13.3M tokens across 39 prompts using claude-opus-5.
Fixed security vulnerability and stabilized test suite
Cetacean
FULL SEND
8.2x your typical session
3.2M tokens vs typical 396.9K
3.2Mtokens
16prompts
3:44:25time
1,976lines
Claude Codeclaude-opus-5Python$224.34
balanced mix
Addressed a security vulnerability and resolved flakiness in integration tests.
@jacob· 16h
Built REDDIT_MARKETING
REDDIT_MARKETING
DEEP WORK
Biggest session this week
Previous best: 1.0M · +1743%
19.0Mtokens
21prompts
22:22:40time
8,580lines
Claude Codeclaude-opus-5TypeScript$272.27
22h 22m session creating new files and components. 18.9M tokens across 21 prompts using claude-opus-5.
Refined energy-mood check-in with Wren's scene-specific posture responses
unnamed-bipolar
DEEP WORK
Most prompts this week
11 prompts
3.5Mtokens
11prompts
2:06:43time
1,302lines
Claude Codeclaude-fable-5-1HTML$129.48
Refined check-in scenes where Wren responds to energy through posture and made the visual control primary.
Shipped interactive animated avatar for the health app
unnamed-bipolar
DEEP WORK
First session on unnamed-bipolar
First build tracked for this project
7.3Mtokens
5prompts
3:39:47time
1,408lines
Claude Codeclaude-fable-5-1Python$192.64
Built an animated avatar character that responds to tap input with smooth state transitions.
Debugged cetacean
Cetacean
Active every day this month
30 consecutive days
420.5ktokens
4prompts
7:59:40time
380lines
Claude Codeclaude-opus-5Markdown$13.58
mostly building
7h 59m session running commands and testing. 420K tokens across 4 prompts using claude-opus-5.
Dirigible
Voxel based colony sim/god game.
2.3Btokens
634
builds
7.2k
prompts
1239.6h
build time
Designed the tracking data model for mood and symptom capture
unnamed-bipolar
FULL SEND
5.4x your typical session
1.8M tokens vs typical 332.3K
1.8Mtokens
3prompts
1:19:26time
1,228lines
Claude Codeclaude-fable-5-1Python$50.67
Documented how mood and symptom data is structured with full design reasoning.
Shipped API fixes for feed hosting and content negotiation
Cetacean
MARATHON
16.2x your typical session
6.4M tokens vs typical 397.1K
6.4Mtokens
28prompts
5:55:48time
3,213lines
Claude Codeclaude-opus-5Go$195.12
mostly delegating
Merged two API fixes: single-host feed routing and per-endpoint content negotiation.
Added CSV export and optimized list performance
Cetacean
FULL SEND
5.1x your typical session
2.0M tokens vs typical 397.1K
2.0Mtokens
7prompts
45:08time
1,753lines
Claude Codeclaude-opus-5Go$83.13
mostly delegating
Shipped CSV export for list endpoints with improved performance and consolidated routing logic.
Fixed 16 product defects and cleared test quarantines
Cetacean
FULL SEND
Top 14% session this week
930 sessions tracked this week across all builders
1.4Mtokens
4prompts
4:16:52time
60lines
Claude Codeclaude-opus-5Go$143.53
mostly building
Fixed 16 product defects end-to-end, cleared all test quarantines, full suite passing.
Verified comprehensive e2e coverage and documented contract testing roadmap
Cetacean
FULL SEND
Top 15% session this week
930 sessions tracked this week across all builders
1.3Mtokens
4prompts
1:25:45time
4,769lines
Claude Codeclaude-opus-5Go$73.98
mostly building
Confirmed zero-gap end-to-end test coverage across all routes and identified contract-level testing as the key next priority.
Résumés tell. Promptbook shows.
Every other platform asks builders to describe themselves. We capture what they actually did.
The old way
With Promptbook
Self-written portfolio
Captured build activity
Polished résumé claims
Live sessions from real tools
Static project list
Recent build velocity
Manual screenshots
Server-rendered build cards
Guess who's active
Search by recency, tools, tokens
How it works
01
Builders connect their tools
One install hooks Claude Code or Codex. Every coding session is captured automatically — no uploads, no self-reporting.
02
Sessions become proof
Tokens, builds, hours, tools and languages turn into a searchable profile and shareable build cards. Never their code or prompts.
03
Companies search the real ones
Filter by what builders actually ship — then reach out. Browsing is free; you pay only to contact.
For companies & investors
Find builders by proof, not promises. Search the full pool free — pay only when you want to reach out.
Browse builders →For builders
Get found for the AI work you're already doing. Free, forever — builders never pay for visibility, rank, or placement.
Get found →Captured live
Streamed from real coding sessions. Nothing self-reported, nothing to upload.
Private by default
Never your source code or prompts. You control which projects are visible.
Tokens, not talk
The honest measure of real AI building — the hardest stat to game.
The database of verified AI builders.
0 builders and counting, 6 open to opportunities.