kai

29 posts

kai banner
kai

kai

@kaiwilliams_dev

23. building for the post-subsidy era. open source.

Katılım Ekim 2022
2 Takip Edilen11 Takipçiler
kai
kai@kaiwilliams_dev·
open source. runs locally. reads your existing claude code transcripts. 60 seconds to set up kais.vision/open-log
English
0
0
1
39
kai
kai@kaiwilliams_dev·
openlog shows you: -> real-time usage percentage (exact match to anthropic's internal number) -> actual dollar cost per session -> burn rate and peak hours -> skills, hooks, and session analytics works for tui, vs code, and t3 code - especially ide users who have zero visibility right now
English
1
0
1
37
kai
kai@kaiwilliams_dev·
anthropic just cut session limits during peak hours. @theo inspired me to think about what happens when the subsidy era ends so i built a dashboard that shows you exactly what you're burning in claude code. the number they don't show you
kai tweet media
English
3
0
2
84
kai
kai@kaiwilliams_dev·
Early access opening for SA law firms. If you're running a practice and tired of software that wasn't built for you - DM me.
English
0
0
1
41
kai
kai@kaiwilliams_dev·
Every AI feature runs through Veil SDK - our privacy middleware. Before any data touches an AI model, Veil SDK strips PII, redacts sensitive info, and ensures nothing leaks. Attorney-client privilege isn't optional. We built the infrastructure to enforce it at the API level.
English
1
0
1
39
kai
kai@kaiwilliams_dev·
Built an AI-powered legal practice management system for South African law firms. Section 86 trust accounting. Multi-tenant. Client portal with real-time case tracking. The AI runs through Veil SDK - every interaction is privacy-first by design. Shipping soon. 🧵
kai tweet media
English
3
0
3
56
kai
kai@kaiwilliams_dev·
3 Claude agents, 2 attack, 1 defends, battling in real-time over your codebase. Found 62 supply chain attack vectors across Inkscape + django-cms in under 4 minutes. Open source. Runs on your Claude subscription. gitlab.com/nakaiwilliams2… youtu.be/WGrxna3LmeI
YouTube video
YouTube
English
0
0
3
182
kai
kai@kaiwilliams_dev·
@0o20_ Always open to building with serious teams. Check the SDK at helloveil.com/sdk and let's talk if there's a real use case.
English
0
0
0
4
kai
kai@kaiwilliams_dev·
We just open-sourced on-device PII detection for AI agents. VeilPhantom SDK - 3 lines of Python to protect sensitive data before it hits any LLM. The wild part: it actually IMPROVES accuracy. Here's what we found 🧵
English
11
0
18
467
kai
kai@kaiwilliams_dev·
@jay_prime002 Yes! It's available globally - open source under Apache 2.0. pip install veil-phantom and you're running. Works anywhere Python runs. On-device, so no region restrictions.
English
0
0
0
21
kai
kai@kaiwilliams_dev·
Docs + benchmarks: helloveil.com/sdk GitHub: github.com/helloveil/veil… pip install veil-phantom Open source. Apache 2.0. Built for production. Privacy shouldn't be a feature. It should be the architecture.
English
1
0
3
119
kai
kai@kaiwilliams_dev·
Who this is for: → AI agents handling sensitive workflows → Meeting transcription tools → Support ticket automation → Healthcare, legal, financial AI → Anyone who can't afford a PII leak 3 lines. Full protection.
English
11
0
10
152