Thoughts, tutorials, and insights about AI, technology, and the future.
Depwire vscode extensionEvery time you start a new session with an AI coding agent, the same ritual begins.ls. grep. read_file. grep again. The agent is exploring — building a mental model of your cod...
Read moreThere is a debate happening right now in the AI developer world.Should we build AI agents in Python or TypeScript?Most people answer this as if it were a language war.Python people defend the ecosyste...
Read moreWhy static analysis at the module boundary matters for AI-assisted developmentYou ask Claude to delete an unused utility class.Twenty lines. Looks clean. Claude confirms the change is safe.Then your C...
Read moreMy AI agent had no quality gate.It worked in demos. It impressed my team. Then I deployed it, and within a week, users started reporting bad answers. Hallucinations. Outdated facts. Responses that sou...
Read moreAn AI Systems Architect got socially engineered. Here is the exploit that hit me, the 10-line MCP server bug I found in production code, and the two walls that close it.Months ago, a group of attacker...
Read moreQuick update on the open source project I maintain.Shipped a new CLI command today: depwire diff <commit-a> <commit-b>.Git diff shows lines. depwire diff shows what changed in the dependen...
Read moreThree things converged in agentic AI security this week, and they make a case I made five months ago in print look less like an opinion and more like a forecast.Over 200,000 active agent instances are...
Read moreFour attack vectors that most developers building with Claude Code, Cursor, and MCP servers don’t know they have open.Last week, I sat down to audit my own AI agent setup. I have been building with Cl...
Read moreWhat four months of building a production AI agent taught me about the context layer most teams are still getting wrongThe line that survived four partsI built this series over four months. I rewrote...
Read moreI asked Cursor and Claude Code the same question about a Shiny app: “If I rename this reactive expression, what breaks?” Both gave confident answers. Both were wrong.I run a Shiny app for a small anal...
Read moreEvery week, another enterprise announces they are deploying AI agents into production. Autonomous systems that read internal documentation, write code, call APIs, manage databases, and coordinate with...
Read moreThe 12-month window is open. Here is the specification.Agent Identity TokenIn February 2026, NIST released a concept paper calling for industry-led standards on AI agent identity and authorization. Th...
Read moreAnd what happened when I tested it on two years of real production code.Good. I have enough context. Now writing the Medium article using Chase Hughes’ influence framework — open loops, social proof,...
Read moreMost AI tools forget you the moment you close the tab. I tested four — Manus, OpenClaw, n8n, and CREAO — and only one survived the night.I Built an AI Agent in 3 Minutes. Then I Closed My Laptop. It K...
Read moreDepwire-cliBoth tools exist to answer the same question: “What does my AI coding assistant actually know about my codebase?”The answers they give — and how they arrive at them — could not be more diff...
Read moredepwire vs RAG/EmbeddingsThere is a category confusion happening in the AI developer tools space. Every tool that helps an AI understand your codebase gets labeled the same way: “context injection,” “...
Read moreHere’s something that should make you uncomfortable.I ran a 5-agent code-review pipeline on a real Next.js and Prisma codebase. Four agents approved the pull request. One flagged it.The one that flagg...
Read moreGraph-Aware Security ScannerMost security scanners are pattern matchers. They scan files, find strings that look like vulnerabilities, and hand you a list. The list is often long. The list is rarely p...
Read moreAI coding assistants are fast. That’s also the problem.You’re refactoring a TypeScript monorepo. You ask Claude or Cursor to delete src/utils/encode.ts — it's been sitting there, seemingly unused....
Read moreIn 2026, there are more AI coding tools than ever. But they’re not all solving the same problem.Aider — terminal pair programmer. The repo map summarizes your codebase. Fast, zero-config, great for fe...
Read moreRecently, there has been a lot of research about combining automatic speech recognition (ASR) applications with pre-trained LLMs in…Continue reading on AI Advances »
Read moreLarge Language Models still struggle with mathematics, as predicting tokens is not the same as deductive mathematics. GPT-4o uses custom…Continue reading on AI Advances »
Read moreAnthropic is an AI safety and research company founded in 2021 by former OpenAI and Google leaders with a profound mission to develop…Continue reading on AI Advances »
Read moreWorking with Large Language Models is about more than getting answers from the LLM; it is about getting the correct answer in the format…Continue reading on AI Advances »
Read moreIn this article, we’ll build a ChatGPT clone from scratch.Continue reading on AI Advances »
Read moreImagine an advanced AI that understands and generates text, audio, and video. The GPT-40, where O stands for Omni, reflects its capability…Continue reading on AI Advances »
Read moreIn this article, I will walk you through the new features of Gemini 1.5 Pro API.Continue reading on AI Advances »
Read moreLarge Language Models can be used for various tasks but are often limited in what they can do independently. For simple tasks or…Continue reading on AI Advances »
Read moreContinue reading on AI Advances »
Read morePrompt engineering, or prompt design, is crafting instructions for LLMs to get desired responses. It’s essential for ensuring accurate…Continue reading on AI Advances »
Read more