Prashant Prakash

Prashant Prakash

Building audit-ready AI systems
#39641701 follower0 following
☀️ 2 day streak
>10,000All time
7 KP

About

Building tools to test whether AI systems can actually justify their decisions. Focused on: - determinism - admissibility - execution boundaries If a system can’t survive an audit, it won’t scale or be trusted.

Badges

Tastemaker
Tastemaker
Gone streaking
Gone streaking
Gone streaking 5
Gone streaking 5

Maker History

Forums

OneCOp/oneco•

16h ago

What should an AI Employee be trusted to own?

AI agents are getting better at completing individual tasks. But we think the more interesting question is what happens when a company has dozens or hundreds of AI Employees doing real work.

At OneCo, we re exploring this model:

AI Employee Skills Tools/MCPs Project Workflow Outcome Human Approval Governance Audit

The idea is not just to give an AI a prompt and get an answer. An AI Employee should have a defined role, access to the right tools and context, a measurable outcome, and clear boundaries around what it can do.

Can Your AI Survive an Audit?p/can-your-ai-survive-an-audit•

5mo ago

Is your AI system actually allowed to act?

Most teams focus on:

  • accuracy

  • performance

  • latency

  • logs

But ignore one question:

Was the system allowed to act on that state at that moment?

•

5mo ago

Can Your AI Survive an Audit? - Test if your AI decisions are actually defensible

Most AI systems work. Very few survive an audit. Test if your AI can defend its decisions — not just produce them. Evaluates: replayability determinism authority boundaries state validity Get: PASS / FAIL Risk classification Failure points + fix path PDF audit report If your system can’t justify execution, it won’t scale or be trusted.
View more