LLMs

When code gets cheap, the hard part is deciding what “correct” means and building a reliable way to check it.

A high-level approach for building AI products from the “outside in”

Coding agents dominate the conversation, but the harness your agent needs depends on its action and context complexity. Here’s how to build for the job in front of you.

Research-backed cues to detect LLM-generated text along with the mathematical intuition as to 'why'

Most coding agents treat prompt construction like retrieval: gather more files, add more context, hope the model figures it out. But that approach breaks down fast. As context grows, irrelevant code competes for attention, and when the window fills, agents start compressing their own memory—often mid-task. What looks like “forgetting” is usually just degraded context. This article explores a different approach: treating prompt construction like a compiler that decides what to keep, what to reduce, and what to discard entirely.

A step-by-step guide to building, running, and monitoring a stateful customer support agent using Python, LangGraph, and Langfuse.

One of the patterns I've found particularly helpful with LLMs is the diverge-then-converge pattern. LLMs excel at both of those things separately, and when we put them together, they are an excellent tool to help us solve problems.

AI tools accelerate output, but human engineers still need better structures, roles, and workflows to avoid cognitive exhaustion.

How post-training transformed language models with raw intelligence into the AI assistants used by billions of people today

Open source voice cloning. Local-first. Free forever.

Handy is a cross platform, open-source, speech-to-text application for your computer

Jake Van Clief, giving you the Cliff notes on the new AI age.

MCP servers are powerful: they let clients (IDEs, agents) interact with remote or local features and...

Thinking Machines' first open weights model is a 975 billion parameter alternative to Chinese LLMs

The seven layers that survive production, and the open source pick for each

Interpretability research on Claude's internal thoughts

AI-assisted development has made frontend work feel much faster. A developer can ask for a form, a dashboard card, a table, a modal, or a responsive layout and

Learn why vague prompts create hidden rework, how acceptance criteria help, and when BDD-style tests are worth the upfront effort.

Agent-native applications are apps built so humans and AI agents can operate the same product, with the same underlying actions, data, and permissions.

Built-in telemetry and OpenTelemetry tracing across every agent lifecycle transition.

AI can handle messy SRE incidents, but only if we give it strict guardrails and keep humans in the driver's seat when things go wrong.

The org-design problem hiding under your agent mesh

In 2025, if you weren't building with MCP, you weren't serious about agents. The Model Context Protocol dominated the agent conversation for the better part of

The fourth article in my context management trilogy

The following article originally appeared on Addy Osmani’s blog site and is being republished here with the author’s permission.Coding agents are

MCP has gained a huge amount of popularity. And for good reason, it's easy to use and can allow you to really harness the benefits of LLMs. So here's a quick MCP tip! Dynamically create MCP servers with FastMCP by passing a REST API's OpenAPI schema to

My notes from the Open-Source LLM Builders Summit in Lausanne.

In addition to using AI in Obsidian, I'm also allergic to voice-to-text.

MCP Tool Poisoning on the main website for The OWASP Foundation. OWASP is a nonprofit foundation that works to improve the security of software.