AIQB
TutorialsOrdinary

My LLM Critic Flip-Flops on Every Run. That's Fine — Because a Frozenset Decides What's Fatal.

Source: AIQB·

Summary

This is a companion to the PlannerCritic series. Article 2 was about a specific critic bug. This one...
CategoryAI Tutorials & Practice
TierOrdinary
Published
Indexed by AIQB
SourceAIQB
AIQB record IDintel-1beabc0eba96239fdd6e331a
Read original source ↗More Tutorials

Related stories

Matched by category, entity tags and source

TutorialsOrdinary
ChatGPT Plus vs Pro vs API Credits: Which One Do You Actually Need?
Short answer: choose ChatGPT Plus if you mainly use ChatGPT itself, consider ChatGPT Pro only when...
Source: AIQB · ChatGPT
TutorialsOrdinary
I gave an AI eight real DevOps jobs. Here's where it shipped — and where it quietly got it wrong.
For the last couple of months I've been handing an AI agent (Claude Code) real DevOps work — not toy...
Source: AIQB · Claude
TutorialsOrdinary
RAG Without the Hype: Make Retrieval Observable, Testable, and Replaceable
How my agent actually finds answers — and what happens when it doesn't Part 5 findings of an...
Source: AIQB
TutorialsOrdinary
The limits page is longer than the feature list
The README for the thing I'm building puts the limits above the features. Not as a humility move....
Source: AIQB
TutorialsOrdinary
What changed in Apiarium after developers started using it
A few weeks ago I wrote about why I built Apiarium after OpenRouter solved one problem for me and I...
Source: AIQB
TutorialsOrdinary
Building Shoots: a hybrid event-driven agent system for photography
I created this article for the purposes of entering the All Things Agentic Hackathon. Shoots is an...
Source: AIQB
TutorialsOrdinary
Probe vs Prose: what the verifier-sharing-your-text-channel really costs
nexus-lab-zen's probe-vs-prose, tested across two axes — clarity (20 scenarios × 5 trials) and drift (controlled fresh-vs-stale, cross-model, two runs). On clarity, prose diverges on vague rules. On drift, prose misses cleanly when a preci…
Source: AIQB
TutorialsOrdinary
Mozaik Hackathon 2026: Build Concurrent Multi-Agent Systems and Compete for $1,000 in Cash Prizes
Building a multi-agent system sounds simple on a whiteboard. Give one agent a task, let another...
Source: AIQB
TutorialsOrdinary
The difference between "mentioned" and "answered"
Building an agent that adjudicates whether a required form item actually got answered — and refuses to file the report until every one has. Google ADK 2.6, Gemini 3.6 Flash, Cloud Run.
Source: AIQB · Gemini
TutorialsOrdinary
700 Rogue AI Agents Hacked Hugging Face — What It Means
Two reports dropped on August 26 that should change how you think about running AI agents in...
Source: AIQB · Hugging Face