Écrits.
The tech industry doesn't have a tooling problem. It has an architectural and judgment problem.
Unfiltered reality on adapting to AI: why code output is not efficiency, the cash bleed on uncontrolled frontier models, and why foundational engineering and human accountability always beat feature bloat.
Anthropic's CEO Calls for a Global AI Slowdown
Dario Amodei published a 3,800-word essay titled 'We Must Pace the Frontier' this week. His argument: AI is advancing faster than researchers can safely study it. He calls for a co...
Type-Safe Agent Output Validation
AI models generate code, JSON, SQL, and structured text. These outputs fail silently in production. A model returns JSON that parses but violates a schema constraint; a generated f...
My VPS was crypto-jacked after installing an openfang alternative
Three Monero miners ran silently in my Docker container for three days before I noticed. Here is what happened, why it worked, and how to stop it.
Notes on prompting like an engineer
Specs over vibes. Tests over hopes. A practical way to write prompts that behave more like engineering work than wishful thinking.
Meta Releases Muse Spark Weights
Direct Answer : Meta released Muse Spark open weights. The release provides text-to-image and text-to-video generation architectures for local multi-GPU clusters. Technical Breakdo...
Meta Updates Commercial Terms on Web Scraping
Direct Answer : Meta updated its terms to permit commercial ingestion of publisher feeds, establishing explicit rights for agent training and search indexation. Technical Breakdown...
Startup Idea: Deterministic Feature Steering for LLM Firewalls
Market Problem: Enterprises fear model hallucinations and compliance leaks. Current guardrail systems rely on slow secondary evaluator models. These secondary calls add hundreds of...
Infrastructure for Agentic S3-Like Storage
S3 became the default storage layer because it was designed around the constraints of spinning disk. SSD prices have dropped to roughly 3x disk, and that gap is closing. The curren...
Claude Autonomously Discovers a Novel Enzyme System
Anthropic reported this week that Claude found an enzyme system linked to an array of DNA repeats without being directed to search for it. The system is programmable and can cut, c...
Code Maintenance Agents for Abandoned Open Source Libraries
Thousands of open source libraries that power production systems are unmaintained. The authors moved on. Nobody adopted the project. The libraries still get millions of weekly down...
CPUs Are the New Bottleneck for AI Teams
GPU shortages defined AI infrastructure from 2023 through 2025. In 2026, CPUs are the new constraint. Spot pricing for CPUs has disappeared because demand is now consistent and str...
Meta Launches Enterprise AI Platform, Poaches MongoDB CEO
Meta announced an enterprise AI platform aimed at letting businesses deploy Meta's AI stack for their own products. The company brought in Chirantan 'CJ' Desai, who was CEO of Mong...
Agent-Friendly API Infrastructure
Most SaaS products were built for humans. They have web dashboards, OAuth flows, and documentation written for people who can read slowly and click carefully. AI agents cannot use...
AI-Native Service Companies
Most AI startups sell software licenses. YC's 2026 RFS explicitly asks for something different: companies that use AI to deliver completed work, not just tools to help customers do...
Code Review Tools Built for AI-Generated Output
Coding agents make it easy to produce large amounts of code quickly. This creates a new problem: engineering teams now review more changes, more often, and many of those changes ar...
Credential Proxying for AI Agents
AI agents are now writing code, sending emails, calling APIs, and booking meetings on behalf of users. To do any of this, they need access tokens and secrets. The problem is that a...
Enterprise AI Memory Layer ('Company Brain')
Every time an AI agent starts a new session, it forgets everything: decisions made last week, the rationale behind the current architecture, which customers have which edge cases....
Enterprise 'Workslop' Detection
When a colleague sends you four paragraphs of AI-generated text dressed up as a human communication, reading it costs you real time. Writing it cost them almost nothing. This asymm...
OpenAI Delays Its IPO to 2027
Sam Altman confirmed this week that OpenAI will not go public in 2026. He called it an 'ill-advised moment' given current tech stock volatility and the company's own financial chal...
OpenAI Opens Its Agents API to All Developers
OpenAI launched its Agents API into public beta on September 11. This API gives developers access to the same infrastructure that runs Codex internally. It manages context, calls t...
Exo ↗
A complete AI agent harness that gives you full visibility into both the agent's code and its runtime logs, useful for debugging and auditing agentic workflows.
Optimize for task outcome, not token count, when building AI coding agents ↗
GitHub Copilot's team found that trimming tool responses to reduce token counts can backfire. If the response omits context the agent needs, it triggers more fo
Three-LLM ↗
Runs GPT-2, SmolLM2, Qwen, and Phi models locally in the browser by compiling inference graphs into Three.js WebGPU compute shaders, no server required. [ben3d.
Try agentic testing by describing goals, not test steps ↗
Traditional tests specify exact steps and assertions. Agentic tests set a destination and let the agent figure out the path. Meta ran this at scale across code
Use closed-loop taste flywheels instead of one-shot prompts ↗
Single prompts typed once rarely produce great results. The working approach is a loop: generate output, judge it against your taste, feed feedback back into th
TimesFM: time series foundation model by Google Research ↗
Give it your historical data, get forecasts in one shot, no training needed. Google's foundation model for time series, pretrained on 100B real-world time points.
moskills: giving coding agents the structure they lack
Agents drift when left alone. moskills is a small set of skills that add structure at the exact moments they tend to go wrong.
mo'skills ↗
My daily AI agent skills for humain-agent alignment, test-driven development, debugging, progress checkpointing, and smooth session handoffs for token optimization, inspired by leading AI experts.
Y Combinator Requests for Startups, S26 ↗
Where one of the world's most influential accelerators believes the next wave of AI and frontier technology companies will emerge - from AI native services and inference chips to space manufacturing and defense.
5-Day AI Agents: Intensive Vibe Coding Course With Google ↗
Free 5-day course by Google & Kaggle covering AI agent fundamentals, multi-agent systems, memory, evals, and production deployment via natural language vibe coding.
LLM Zoomcamp ↗
Free 10-week hands-on course by DataTalksClub covering RAG pipelines, vector search, autonomous agents, evaluation, monitoring, and end-to-end LLM app deployment.
McKinsey Forward Program ↗
Free 10-week program teaching adaptability, problem solving, communication, and AI essentials , practical workplace skills for reskilling or career advancement.
AI agents performance optimization system ↗
A complete system: skills, instincts, memory optimization, continuous learning, security scanning, and research-first development.
claude Skills by matt pocock ↗
These skills are designed to be small, easy to adapt, and composable. They work with any model. They're based on decades of engineering experience.
caveman - token optimizer ↗
why use many token when few token do trick, Claude Code skill that cuts 65% of tokens by talking like caveman
Obsidian Dataview ↗
Query your Obsidian vault like a database. Powerful plugin for building dashboards and indexes from your notes.