Archive
Browse posts by year and month
2026
91 postsOctober 4 posts
September 25 posts
- MCP Agent Verification That Proves Every Claim's Source
- Nemotron 3 Diarization Flips the Speaker Labeling Math
- OpenAI Proaction Case Study Decoded, Claim by Claim
- Is a Consumer AI Agent Worth It? $550 Saved, $64 Wasted
- AI Agents vs Workflows? Count Your Branches First
- Pass@k Crossover Tests Tell You When to Ship the Base Model
- Googlebook for AI Development, Judged by Two Numbers
- Entity Deduplication From 50 Million Pairs to Thousands
- Qwen3.8-Omni-Flash vs Gemini Flash Real Cost per Task
- The Claude OpenAI Security Incident, Decoded for Builders
- Gemini 3.8 Live Extended Thinking Needs a Latency Budget
- LLM Prompt Caching Can Cut Input Costs Up to 90%
- Open Weights vs Frontier Models Is a 4.4 Month Call
- Async GRPO Training on Serverless GPUs Cuts RL Cost
- AI Agent Monitoring Beyond the Dashboard
- Devin GPT-6 Astra Self-Testing Moves Review to Evidence
- Training a Diffusion Model From Scratch in 3.5 Days
- OpenAI Navier-Stokes, a Reported $40M Lesson in Verification
- Netflix MAPS Teardown and What Actually Transfers
- Meta's AI Second Brain Runs on a Knowledge Supply Chain
- AI Text Watermarking in Python, Three Families Tested
- KV Cache Math for Million Token Agent Runs
- GPT-6 Astra Pricing at $6 an Hour, Audited
- Prompt Dependency Graphs That Shrink Your Retest Set
- TontaubeV1 Review With Serving Math You Can Run
August 25 posts
- ChatGPT DSA Designation Sweeps In Every AI Search Engine
- oMLX Review, Auditing the 90s to 5s Agent Claim
- NLP vs LLM vs RAG, Routed by Task Shape and Cost
- NVIDIA Hugging Face Acquisition Ends the Neutral Hub
- Text-to-SQL Evaluation That Catches Silent Wrong Answers
- MoE Serving Cost Math for 6 of 125B Active Parameters
- AI Agent Memory Lessons From LinkedIn's Hiring Assistant
- Why Speculative Decoding Pays Nearly 4x on CPUs
- AI Agent Token Usage Overtook Humans on OpenRouter
- How WhatsApp Scam Alert Detects Scams It Cannot Read
- AI-Assisted Product Launch Teardown of Stampli's 68% Claim
- Build an AI Text Detector, Then Test If It Can Ship
- The Real Cost of Gradient Accumulation on T4 and L4
- LLM Context Window Management With a Token Budget
- The Real OpenAI Ultrafast Mode Speedup, Workload by Workload
- LangChain vs LangGraph for Stateful AI Agent Orchestration
- ChatGPT Business Premium Pricing Decodes Agent Token Math
- Muse Glimmer Local Review Tests 30B Agents Under 20GB
- Structured Output Local LLM Tactics That Survive Production
- AI Agent Cyber Security Evaluation After the Astra Slowdown
- Agentic Loop Token Costs Are an Architecture Problem
- Megakernels in LLM Inference When Fusion Actually Wins
- Stacked Pull Requests Relocate AI Mega-PR Review Cost
- Disaggregated GPU Inference Hits the KV Cache Wall
- How The Copilot Prompt Injection Worm Spreads
July 28 posts
- LLM-Native Recommendation Architecture After Netflix GenRec
- How MCP Servers for AI Agents Bridge Fragmented Data
- Control Reasoning Effort LLM APIs in Production
- BM25 vs Dense Retrieval and SPLADE for Production RAG
- Shipping LLM Browser Agents Without Breaking Production
- Claude Opus 5 Prompt Injection Hits 0% Across 129 Tests
- AI Coding Agents Cut Per-Task Cost Up to 4x With Indexing
- ChatGPT Shared Link Vulnerability Plants Rogue Agents
- Grok vs Copilot in Excel for Builders
- Run AI Models Locally on Mac With MLX and Nativ
- Hugging Face Agentic Attack Redefines AI Security Response
- AI Agent Cost Per Resolution Decides If It Ships
- Self-Host LLM Inference, the Netflix Decision Framework
- Where AI in Media Production Workflows Actually Pays Off
- Orchestrate Open Source LLMs vs Frontier Models
- OpenAI Codex Subagent Encryption Breaks Agent Observability
- ChatGPT Sales Workflows Fail on Real CRM Data
- Real-Time AI Dental Image Verification Cuts Claim Denials
- Fine-Tuning vs RAG vs Prompt Engineering Decision Framework
- How LLM Agent Scaffolding Fixes Failing Code Review Agents
- Building Proactive AI Agents With Context Graphs
- AI Coding Benchmarks Are Gameable by Design
- AI Video Editor Limitations Break Iterative Workflows
- LLM Vendor Data Risk Has a Break-Even Price
- Stop RAG Hallucination with Typed Schema Contracts
- Domain-Specific LLM Evaluation Demands Leakage-Free Data
- Why an Agent-to-Agent Gateway Beats Point-to-Point Links
- LLM API Token Inflation Hides Claude Sonnet's Real Cost
June 9 posts
- Production LLM Application Security Beyond Prompt Injection
- Secure AI Coding Agents From Supply Chain Attacks
- Build Expert-in-the-Loop AI for Reliable Automation
- Build an LLM Fallback Strategy Before Access Drops
- Human-in-the-Loop Systems: Balancing AI Speed and Operational Safety
- Custom AI Inference Chips Are Eating the GPU Market
- The MLOps Playbook: Choosing Cloud GPU Providers for LLM Inference
- The End of AI Reasoning Transparency: When Chain of Thought Becomes a Summary
- Why AI Agents Leak Sensitive Data (and How to Stop Them)