The blog.
23 posts on ai-agents, page 1 of 2RSS feed
Claude Duet: What It Is and How to Get Access
Anthropic's Claude Duet is an always-on personal AI agent with its own virtual computer and browser, powered by Opus 5.5. Here's how it works, who it's for, and how to get on the waitlist.
Claude Projects: Cloud Multi-Agent Orchestration
Claude projects turns Claude into a cloud project manager that runs a fleet of full Claude sessions 24/7. Here's how to get access, when to use it, and how it beats sub-agents.
Claude Mods: The Biggest Claude Code Upgrade Yet
Claude Mods let you rewrite, customize, and extend Claude Code itself. Here's how they work, two real mods in action, and the one audit prompt that builds your own.
ChatGPT Dots: Early Access Review and How It Works
I got early access to OpenAI's new Dots — an always-on cloud AI agent powered by GPT-6 Astra. Here's how it works, how it compares to Muse and Grokbot, and who actually needs it.
How to Build an Agentic OS (Claude OS) That Actually Works
An Agentic OS isn't about fancy dashboards. It's a three-layer system — skills, memory, and visual — that routes your work across GPT-6, Claude Opus 5.5, and Jev so AI actually takes work off your plate.
5 GPT-6 Astra Mistakes Killing Your Results (And Fixes)
GPT-6 Astra is the best model we've seen, but most people are hamstringing it. Here are the five mistakes killing your Astra results and exactly how to fix each one.
10 New GitHub Repos Every Claude Code User Should Try
A rundown of 10 brand-new open-source repos worth adding to your AI dev stack right now — from the DeepSeek Harness and OmniRoute to Orca, Herdr, and Firecrawl's Anydoc.
Claudex Loop: Stop Letting Claude Grade Its Own Work
Claude grades its own work too favorably. Claudex Loop fixes that by bringing Codex in as a second set of eyes on both the plan and the code, across four phases.
10 Claude Code Concepts That Actually Matter
The 10 Claude Code concepts that actually move the needle — from picking the right model and effort level to loop engineering and dynamic workflows.
Grok Bot Review: Why It's a Waste of Time
Grok Bot is xAI's new team-of-agents tool, and it costs $200 a month to use the feature everyone's hyping. If you already know Claude Code or Codex, here's why it's not worth your time.
The Claude Code Guide: What Actually Matters in 2026
A no-BS roadmap for mastering Claude Code in 2026 — from the desktop app and prompting to skills, MCPs, loop engineering, and building your own agentic OS.
Graph Engineering vs Loop Engineering: What Changed
Graph engineering is loop engineering split across multiple connected agents, each running its own loop with its own success criteria. Here are the three scenarios where it is worth the complexity.
