ICME Labs Blog

Sign in Subscribe

Latest — Jul 15, 2026

Human-in-the-Loop Alternatives: How to Keep Control of AI Agents Without Approving Every Action

Human-in-the-loop doesn't scale; it decays into a rubber stamp. Here are the five real alternatives, why only one of them actually makes decisions, and how deterministic, provable authorization lets agents run at machine speed without running unsupervised.

More issues

Are You a Wayfinder?

With a debt to Michael Arrington's 2010 essay, "Are You A Pirate?" This is the same fire, carried in a different canoe. Most people are risk averse. That's not an insult, it's just economics. If you want a normal person to take

Neurosymbolic approaches to natural language formalization: what is state-of-the-art?

Formal methods are genuinely great. Done right, they give you hardware that does not betray you and code that provably does what it says it does. But there is a catch, and it is a big one. The proof is only as good as your specification. You can verify a

LatticeBlindFold: Towards The First Zero-Knowledge Lattice-Based Folding Scheme

Bringing NovaBlindFold's zero-knowledge guarantees into the post-quantum world The Problem The intersection of Zero-Knowledge (ZK) proofs and succinct protocols forms a central and rapidly evolving focus in modern cryptography; enabling parties to prove the validity of a computation without revealing the underlying data, while maintaining

Vericoding: The End of "Trust Me Bro, The AI Wrote It".

92% of developers use AI coding tools daily. Trust in AI-generated code has dropped from 77% to 60%. The gap between those two numbers is where the next billion-dollar problem lives.

Getting Found by Agents: A Builder's Guide to Tool Discovery in 2026

Getting discovered used to be a human problem. You stuffed keywords into title tags, begged for backlinks, wrote blog posts nobody wanted to read, and prayed to the Google algorithm. The audience was a person with ten blue links and a short attention span. You were fighting for a click.

What is Automated Reasoning?

The 400-Year Idea That Now Runs AWS Infra at Billion-Query Scale, With ICME's ZK(ARc) Will Power Secure Agentic Commerce At Billion-Transaction Scale.

AI Agents Can Move Money. Lobstar Wilde Proved They Can Lose It Too.

Every security problem eventually becomes a game. Someone builds a wall, someone else finds the door. Someone writes a rule, someone else learns the language of the rule well enough to satisfy it while breaking everything it was meant to protect. We've played this game with email spam,

Succinctly Verifiable Agentic Guardrails With ZKP Over Automated Reasoning - or how to prevent the Agentic Commerce mega-hack.

Agents often have loosely defined guardrails that depend on ‘observability’ & ‘reputation’ and ultimately a human in-the-loop. ICME's Succinctly Verifiable Automated Reasoning allows for guardrails that deliver powerful protection. Use AI guardrails based on math, not trust.

Trustless Agents can't work without Trustless Agentic Memory.

Your agent just made a million-dollar decision based on 'memories' it pulled from a vector database. A TEE attested that the computation was correct. But can you prove they used the right AI memory? Introducing Kinic-CLI: zkML powered Trustless Agentic Memory (zkTAM).

About

ICME Labs Blog

Blog about cryptography, defensive AI, and zero knowledge machine learning.

Topics

AI

Cryptography

Startups