Every case study, system breakdown, and field note, newest first and filterable by topic. No theory, no demos. Only systems that run in production and what it took to get them there.
This is the complete archive: all 781 posts across 63 topics, newest first, filterable by category. It holds the case studies, the model and tool comparisons, and the field notes from systems I built and run. For the curated view, where the writing sits alongside the free Claude Code skills, start at the Library instead.
The archive leans in four directions: AI Search Visibility (137), AI Agent Architecture (72), AI Business Ideas (41), and Open Source (41). Those four account for most of what I publish, because they are where most of the client questions land.
Three places to start. The AI search visibility guide is the hub for the largest cluster and links out to every spoke in it. Grok Imagine vs Midjourney is the image model comparison, rebuilt against live leaderboard data rather than left to go stale. OutreachIQ is the longest running system breakdown here, from first prototype through to a public repo.
Hermes Mixture of Agents for builders: several reference models, one aggregator, higher cost, better review, and when the routing actually pays for itself.
A practical comparison of cloud GPUs, hosted AI APIs, and home AI hardware for local models: cost, privacy, latency, maintenance, electricity, and when a hybrid setup wins.
A practical local AI hardware guide with prices and buy links for Ollama, LM Studio, Qwen, Gemma, Llama, DeepSeek, GLM, Mac mini, Mac Studio, RTX PCs, DGX Spark, and cloud GPUs.
Austin Marchese breaks down the B.U.I.L.D. framework for making Claude Code improve over time. Here is the practical version: knowledge base, bulk ingest, data inflow, review loops, and routines without system drift.
Wes Roth walks through Hermes Agent with Stripe payments, Nous Portal, OpenAI Codex, NVIDIA NemoClaw, and Nemotron 3 Ultra. Here is the practical builder version: agents can run business workflows, but money movement needs hard approval gates.
A practical local AI starter guide inspired by Alex Finn: what local AI is, why it matters, what hardware and software to start with, and which private workflows are worth testing first.
Seven Hermes Agent use cases that hold up in practice: computer control, competitor research, memory, Linear and Notion handoffs, Slack, crons, and skills.
Shashank Agarwal showed Andrew Warner how API.market runs with OpenClaw agents for PR, LinkedIn, seller outreach, finance, hiring, and chief-of-staff work. Here is the practical agent operating system behind it.
Andrew Warner and Matthew Berman break down AI agent loops: triggers, verifiable goals, LLM-as-judge loops, sub-agents, scheduled automations, token budgets, and reusable Loop Library templates.
Alex Finn walked through Hermes Agent's Reach Release: native iMessage, automatic background agents, Unreal MCP, desktop upgrades, profile builder, Skills Hub, smarter memory, and Telegram formatting.
Sakana Fugu and Fugu Ultra are not normal frontier models. They orchestrate multiple models through one API. Here is how they work, what the benchmarks mean, what it costs, and when builders should try it.
SpaceX reportedly buying Cursor is less about one code editor and more about the race to own the general agent workspace. Here is what builders should watch, test, and migrate carefully.
The Fable 5 access pause is a reminder that cloud AI is rented intelligence. Local models give builders a private, offline, resilient fallback layer for everyday work.
Codex Sites, Cursor Canvases, DeepSeek V4 pricing, Poke in Apple Messages, Microsoft Scout, and Hermes Desktop point to the same shift: agent work is moving into super-apps and shareable work surfaces.
Which Hermes Desktop surfaces actually matter, why profiles are not security sandboxes, and what to get right before you schedule anything. Judgment, not docs.
A public-safe prompt for building a local-first AI agent control center with Codex, Claude Code, Cursor, or another coding LLM, without exposing private data.
Codex Sites is not just a one-prompt app builder. The real pattern is memory, safe actions, skills, save-gates, and proof loops that let agents operate internal tools.
OpenAI's Codex role-specific plugins, Sites, and annotations show the next agent shift: packaged workflows for sales, data, creative, product, finance, and internal apps.
OpenAI is pushing Codex beyond coding with role-specific plugins, Sites, annotations, and a deeper ChatGPT path. Here is what builders and operators should understand.
AI browsers promised to put an assistant beside your tabs. Codex, Claude Code, and Claude Cowork suggest a bigger shift: task-based agent workspaces where the browser, files, tools, and review surface live together.
AI plugins, MCP tools, and work plugins are becoming the workflow layer around AI agents. Here is how businesses should think about permissions, tools, review, and rollout.
AI agent memory is becoming the new business knowledge base. Learn how memory differs from chat history, why agents forget work, and how small teams can structure context for better automation.
Andrej Karpathy joining Anthropic is a signal that the next AI advantage is not just the model, but the context, memory, workflows, and research loops around it.
Codex, Claude Code, and Hermes now share a /goal pattern: durable objectives, evaluators, stopping conditions, and safer long-running agent work.
Every post here is about a system that actually shipped. Book a free call and let's talk about what could ship for you.
Book Free 30-min CallAdd your name and email to unlock recording.
Your recording is emailed to me and kept on this site's server for 12 months, then deleted. Privacy policy.
Thanks for your voice message. I listen to every one myself and I will get back to you within one business day.
Explore João's services and public work, or arrange a free 30-minute consultation.
Choose a time with Cal.com. Your booking status is shown there. No calendar details are shared with this assistant.
Open booking in a new tabWhen enabled, each question is processed by Cloudflare AI and checked for abuse with Turnstile. No chat history is stored by this app. Please leave out private information.
Prepared answers · Public information