Featured
Memory Poisoning: The Agent Attack That Survives a Reset
OWASP ASI06 corrupts an agent's stored state once and it acts on the lie forever. Here's how the attack works and the defenses that actually hold.
Gen α AI · Field notes for AI builders
Depth over hype, for people who bet on AI.
Evidence-first analysis of agentic systems, model evaluation, and the economics of AI software. We read the system card, find the primary source, and tell you what actually changed — and what didn't.
Evidence over vibesDepth over volumeHonest about uncertainty
111Deep dives published
9Evergreen pillar guides
BiweeklyThe field briefing
Editor’s picksNew here? These are the pieces we’d hand you first.
01 · AI Tools
AI Coding Tools in 2026: The Power-User Field Guide
The gap between demo and production is the harness you build around the model, not the model you license.
Pillar guide → 02 · Search & GEOGenerative Engine Optimization: How to Earn AI Citations
Search is becoming synthesis. If ChatGPT, Perplexity, and Google's AI Overviews don't cite you, you're invisible, and…
Pillar guide → 03 · Agents & HarnessesAgent Harness Engineering and Agentic Loops: 2026 Field Guide
Execution loops, externalized state, and verification gates now matter more than raw model IQ. Here's how the agents…
Pillar guide → 04 · AI ToolsAI Coding Agent Economics: Real ROI and Cost per Pull Request
Frontier labs now ship more AI-written code than human-written code, but the viral ROI numbers are wrong. Here is the…
Pillar guide →The latestFresh analysis, published continuously — the full archive lives in the rail and the pillars below.
Security & SafetyAI Safety Routing Is Real. The Audit Trail Isn't Yet
Routing risky prompts to safer models can be a serious governance control, but only if buyers can inspect the classifier, fallback chain, logs, and audit evidence.
12 minJune 21, 2026→
AI EconomicsKV Cache Compression Is the New Inference Lever
The highest-leverage serving work in 2026 is no longer just faster kernels; it is shrinking the cache that long-context models reread on every decode step.
11 minJune 21, 2026→
Agents & HarnessesThe MCP Server Boom Moved the Moat to Gateways
The protocol is becoming boring infrastructure; the hard decisions now live in authorization, isolation, observability, and gateway choice.
11 minJune 21, 2026→
AI FrontiersAI Feature Engineering Is the Product Moat Now
Model choice still matters, but the compounding advantage in AI products is shifting to data shape, retrieval signals, feedback loops, and eval labels.
10 minJune 21, 2026→
Agents & HarnessesVoice Agent Latency Hit a Wall. Design Around It
The best AI voice agents now win on interruption handling, endpointing, ASR recovery, and multilingual switching as much as raw milliseconds.
11 minJune 21, 2026→
AI FrontiersSiri AI Is Now a Routing Problem Developers Own
Apple's WWDC 2026 reset makes Siri a test of routing, App Intents, regional gates, and how far developers can trust outsourced frontier AI.
11 minJune 21, 2026→
Agents & HarnessesAI Agent Identity Is the Next Platform Battle
The winning agent stack will be judged by how cleanly it issues, scopes, meters, audits, and revokes bots.
10 minJune 20, 2026→
AI FrontiersPhysical AI 2026 Hits the Jobsite Bottleneck
XPENG's robotics pivot shows the category has arrived, but site robotics will be won by edge inference, BIM loops, and bounded autonomy before humanoids scale.
12 minJune 20, 2026→
AI ToolsAI Radiology Report Generation Moves Into the Cursor
The radiology AI shift that matters is happening inside dictation and signing workflows, where context, audit, and review already exist.
10 minJune 20, 2026→
AI FrontiersEU AI Act GPAI Enforcement June 2026: 14 Moves Before Fines
A 43-day, evidence-first checklist for startups that ship GPAI models, fine-tunes, wrappers, or AI products into Europe.
11 minJune 20, 2026→
Models & ReleasesThe Fable 5 Mythos 5 Export Directive Hit Your API
The U.S. Did more than pause a model; it turned model access into an availability risk engineers have to design around.
11 minJune 20, 2026→
AI FrontiersAI Inference TCO 2026: Tokens Beat FLOPS
Blackwell, MI400, and Trainium are competing on delivered tokens per watt, software maturity, and power envelopes, not spec-sheet peak math.
11 minJune 20, 2026→
Explore the pillarsNine durable guides that organize everything we publish.
AI Tools15 pieces
AI Coding Tools in 2026: The Power-User Field Guide
The gap between demo and production is the harness you build around the model, not the…
Explore →Search & GEO9 pieces
Generative Engine Optimization: How to Earn AI Citations
Search is becoming synthesis. If ChatGPT, Perplexity, and Google's AI Overviews don't cite…
Explore →Agents & Harnesses17 pieces
Agent Harness Engineering and Agentic Loops: 2026 Field Guide
Execution loops, externalized state, and verification gates now matter more than raw model…
Explore →AI Economics10 pieces
AI Coding Agent Economics: Real ROI and Cost per Pull Request
Frontier labs now ship more AI-written code than human-written code, but the viral ROI…
Explore →Model Evaluation13 pieces
Evaluating AI Models and Agents: The 2026 Field Guide
Why static leaderboards lost authority, and how to build an eval program that survives…
Explore →Memory & Context11 pieces
Context Engineering for AI Agents: Memory, RAG & MCP
Why the context window, not the prompt, is the real bottleneck, and how to engineer…
Explore →Security & Safety8 pieces
Securing AI Agents and LLM Apps: The 2026 Threat Model
Why indirect prompt injection, tool-mediated exfiltration, and rogue agents now define LLM…
Explore →Models & Releases10 pieces
AI Models 2026: The Mid-Year Frontier and Open-Weight Map
How the open-weight cluster closed the gap, why reasoning became the default, and which of…
Explore →AI Frontiers18 pieces
AI Frontiers 2026: Diffusion Models, Multimodal AI & More
A practitioner's map of frontier AI in mid-2026, where independent measurement finally…
Explore →