The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
The string "1,234" can mean different amounts under different number formats. Removing punctuation isn't a safe normalization rule.
My Procys work connects OCR and LLM layout reasoning with deterministic validation and exception review.
For an invoice pipeline, I'd retain the original field text alongside the normalized amount and currency. If the document doesn't establish the number format, that ambiguity should go to review before export.
I want reviewers to be able to trace each normalized amount back to its source.
kapilchauhan.netlify.app
Procys: Document intelligence | Kapil Chauhan
Production LLM agents for document processing and identity verification
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
The fix for an agent that keeps getting your design system wrong isn't a better prompt.
It's putting the rules where the agent reads them.
In the repo. In skills. In checks it can run itself.
Writing a course about exactly this.
What sells it for me is the refusal telling the agent exactly where the token should go. A bare "no hex" error usually just sends it looping. Does guard.mjs run only as an agent hook, or in pre-commit too so it catches humans as well?
Testing two color directions for Pilotly, an AI agent platform for product teams.
Which direction would you pick for an AI automation product?
3 voted
100%
0 voted
0%
3 votes
Closed
I'd pick Violet Desk. Blue is what every SaaS uses, violet stands out and still says AI, and the orange sticky notes pop more against it. Does the hero chip row under the headline work on mobile?
Most Claude Code sessions die in the middle — not because the model is weak, but because the brief was vague and the scope kept growing.
Here's the six-step loop I use so a feature lands as a reviewable PR instead of a sprawling rewrite:
01 BRIEF — write the win condition, in-scope files, and what must not change
02 MAP — explore the call graph; confirm the plan before any edit
03 SLICE — one vertical cut that proves the behavior; defer refactors
04 PROVE — failing test first, then implement until green
05 GUARD — diff the change against the original brief; reject drive-bys
06 LAND — conventional commit + risk note; human still owns merge
If scope drifts, return to MAP. Never keep editing blind.
#Claude #ClaudeCode #AIAutomation
