It Remembers Across Restarts — and Forgets Mid-Conversation
Session files stored fourteen messages while the model saw six. A restart keeps your name, and seven turns later the same conversation forgets it.
Tutorials
AI Search Often Knows a Brand’s Category but Misses Its Customer Fit
Being named in an AI-generated answer is not necessarily a win. A brand can appear prominently while...
My Agents Ran in Parallel. I Didn't
A review queue for people who run many AI agents, and the rule that keeps it safe: the model...
Your AI Agent's Patch Is Fine. The Repo Won't Let You File It.
The first hard failure in an agent-written contribution is not a failing test. It's a closed door,...
Same Agent, Same Prompt, Two Briefs: What Hindsight Memory Changed
I ran the same negotiation agent twice on the same deal. Same model, same prompt, same tool loop. One...
One approval, two payments: what an AI agent does after a crash
Written with help from AI (Claude). I maintain Call4Agents, the library mentioned at the end. Every...
I built an MCP server for glucose data that refuses to lie
Ask an AI assistant for a glucose reading through most integrations and you'll get a number back. 132...
Arabic breaks most prompt templates — here is the structure that survives
Most prompt engineering advice assumes English. You copy a six-part template, translate it into...
Being, Doing, Having: Why 50 Years of Senior Craft Lives in Context, Not Code
Why does standard AI code generation try to jump straight to 'Have' without the proper 'Be'? How ladder rungs, the 'Null-State' couch flash, and stand-up comedy explain what separa...
Reverse Engineering with AI: Your Decompiler Can Lie to You
From serialized keys to candidate writers: why an AI agent still needs machine-code evidence TL;DR....
YouTube Description Updates Were Linked to Higher AI Retrievals in a Small Peec AI Test
A small experiment from Peec AI suggests that improving YouTube video descriptions, including adding...
Localizing a static site into 6 languages at build time with an LLM — and validating every string
A small build-time i18n pipeline for a hand-written HTML site: extract strings, translate them in batches with an LLM, validate placeholders and tags, fail the build on anything mi...
Building an LLM Council: Using Multiple AI Models to Stress-Test Startup and Software Ideas
Introduction A few weeks ago, my team and I were stuck. We’re building a waste management...
Stop asking "is this text malicious?": how I stopped 144/144 prompt-injection attacks on an AI banking agent
An AI agent that reads your e-mail and can send money is a strange kind of software. Half of its...
I ran 3 Claude Code architecture diagram tools on my repo. Here's what each one got right
My project, worklore, is a small app that grew in all directions: a Python backend on AWS Lambda, a...
LLM Returns Invalid JSON Once in a While: Structured Outputs vs Validate-and-Retry
LLM JSON that sometimes fails to parse: why it happens, when constrained decoding fixes it, and a validate-and-retry loop for the rest.
An Update on the Hacktoberfest Open-Source AI Challenge: Ghost in the Machine Winners
Hey all, we have a quick update for everyone who participated in the Hacktoberfest Open-Source AI...
Beyond Vibe Coding: Building Invariant-Driven Agentic Systems with AWS Kiro
This post was written as part of the Kiroween 2026 hackathon. The Vibe Coding...
DSH Have Voice
Your agent works for twenty minutes. You are in another window. It tells you what happened — in its...
Do Parallel Claude Code Sessions Share a Prompt Cache? Not Across Worktrees
Claude Code's prompt cache is scoped to one directory, so each worktree writes its own prefix. A jq query over your transcripts shows the reads and writes.
Anthropic Says Claude Speeds Up Biomolecular Modeling With Less Compute
Anthropic says Claude can substantially accelerate biomolecular modeling, a computational field used...
Ambiguous User Requests: When AI Agents Ask, Act or Escalate
Your agent didn't misunderstand you. It guessed — confidently, silently, and wrong. That's the real...
Closing $23,000 Enterprise Deals at the Speed of Chat: Building RelayOS for Switch
When an enterprise prospect asks for a custom ERP adapter, 1,000 req/min quota, and bespoke pricing...
AI Overviews: Why Brand Mentions Matter More Than Backlinks
Backlinks remain an important part of the web's discovery and authority ecosystem, but they may not...