You already have the agents.
What you're missing is the methodology.
Learn to run your own AI dev team — set up your harness, write the specs, and delegate real work through tickets. 1:1 with a mentor who does it every day, on your own machine and your own repos.
What are AI tools, really?
You were promised 10x. Your agents gave you noise.
The tools change every week and you're always a step behind. But the tools were never the problem — what's missing is a method for putting them to work.
You launch agents with no harness
No CLAUDE.md, no skills, no workflows. The agent guesses at your context and hands back generic output that ignores how your codebase actually works.
No specs, so the output drifts
Without a spec that defines the work before you launch, every run wanders somewhere new and no good result is reproducible.
You can't trust the diff
So you read every line and rewrite it by hand — slower than doing it yourself. The 10x never shows up.
Four pillars that turn your agents into a dev team you run
The tools are ready. The method is what makes them deliver — and it comes down to four things we build with you, 1:1, on your own repos.
Configuration is everything
Your harness — system prompt, CLAUDE.md or AGENTS.md, skills, plugins, and workflows. The right plugins ship them ready to run.
Spec-first: define before you launch
Grill the idea, write the spec, break it into tickets. Cover the points and complete the process before you unleash a single agent.
Delegate to your agent team
Work through tickets, hand real work to your agents, and run your own AI dev team.
Stay current
Follow the news, join a community that onboards newcomers, and know your hardware — its capabilities and its limits.
Your mentor shows you how to spec the work, delegate it to your agents, and review the diff with confidence.
We start with what frustrates you most about your workday
The free intro interview is your first session — a real discovery call, not a sales pitch. From there, the mentorship is built around your actual work, not a generic syllabus.
You share your workflow
Your current repo, stack, and daily tasks. It starts in the free intro interview — your first session with your mentor.
We find the pain points
Together we spot the biggest time sinks and friction — the tedious PR reviews, the flaky tests, the boilerplate you keep rewriting by hand.
We design around YOUR use cases
The mentorship is built on your actual work, not a generic syllabus. Every session targets a real task from your day.
First wins come from what you like least
We automate the tasks you enjoy the least first — so the earliest payoff lands on the work you'd most like to stop doing.
The files that make your agents work like a team
Across your sessions you write the artifacts that drive real work — CLAUDE.md or AGENTS.md, skills, tickets, and workflows — and use them to automate the parts of your job you keep doing by hand.
These files follow industry standards (.agents/skills/{skill}/SKILL.md) and work across every agent CLI — Claude Code, Codex, Cursor, and more — so the method travels with you, not with one tool.
What you'll learn across your 1:1 sessions
Not a mass course. It's 1:1 sessions on your own machine and your own repos — you bring the work, and we build the method around it, one session at a time.
Fundamentals
- •Install and configure your harness
- •Write your first CLAUDE.md
- •Prompting and context basics
- •Drive an agent from the terminal
Integration
- •Add skills and plugins
- •Wire up workflows
- •Connect your editor and CLI
- •Review agent output with confidence
Personalization
- •Spec-first: grill → spec → tickets
- •Custom skills for your stack
- •Context and memory management
- •Automate tests, docs, and refactors
Mastery
- •Delegate real tickets to your agents
- •Run your own AI dev team
- •Optimize your dev loop
- •Stay current: news, community, hardware
Learn from a developer who builds agent teams
Not a theorist. A professional who applies these tools in real projects and teaches you from experience.
Joel Dávila
18 years building software — from senior Java to full-stack (Next.js, Vue). At Matchbox he leads AI Enablement, designing the agent harnesses and teams that Logbook teaches you to run. Anthropic-certified in Claude Code, agent skills, and subagents. Creator of Digiall Tech News, an AI-generated news podcast he produces end-to-end with his own multi-agent pipeline.
Matched to your stack
You work with a mentor who lives in your stack — your languages, your agent CLI, and the workflow you ship with every day.
Live, never recorded
You share your screen, ask questions in the moment, and receive immediate feedback on YOUR work.
The same mentor throughout
You work with the same person who already knows your context and goals — the same mentor throughout your sessions.
Not theory — a product shipped entirely with agents
Digiall Tech News is an AI-generated tech-news podcast your mentor designed and runs end-to-end — with the exact methodology you learn here. New bilingual episodes ship from a repeatable agent pipeline, not by hand.
How an episode builds itself
Research
A web-search agent finds the day's stories and de-dupes against past episodes.
Script
A writer agent drafts exactly five host-voiced stories against the show spec.
Fact-check
A separate model validates every claim — the writer is barred from grading itself.
Voice
Text-to-speech renders two synthetic hosts, in Spanish and English.
Mix & publish
Narration, music, and effects are mixed and the episode is exported automatically.
Built without Claude Code — on purpose
The whole pipeline runs on PI — a coding CLI — extended with a custom extension, guided by an AGENTS.md instruction file, with Codex as the model. Not Claude Code. Same methodology, a different toolchain. That is the proof: what you learn here travels across every agent CLI, not one vendor.
What you will achieve with Logbook
Real professionals, real results
From connecting enterprise systems to reviewing legal cases — our mentees apply AI to solve real problems in their daily work.
I managed to connect my company's ERP to an AI agent. Now I can request reports in natural language and get answers in seconds instead of navigating through complex menus.
I was able to thoroughly review a judicial case and extract all the key points needed for the defense. What used to take me days now takes hours, with much greater accuracy.
Invest in your professional growth
1–2 hours · pay per session, in advance
- 1:1 live mentorship, screen-to-screen
- Work on your real repos and daily tasks
- Your own harness — CLAUDE.md, skills, custom agentic files
- You keep everything you build
- Evenings, after your workday (weekday sessions)
- Access to the Logbook community
Need team training or a talk? See services below.
Scheduling & cancellation
Weekday sessions, evenings from 5:00 PM (CDMX); your recurring slot is set in the free intro. Reschedule up to 48 hours before a session. Sessions cancelled with less than 48 hours’ notice, or missed, are non-refundable and count as used.
Ways to work with me
1:1 mentorship is just the start. Bring your team, or invite me to speak at your next event.
Weekday evenings · remote
Remote 1:1 mentorship on your own machine and repos. Start with the free intro session.
Book a free intro sessionFor multiple people
Training for your whole team, in the format that fits:
- •Workshops
- •On-site Saturdays
- •Classes
- •Weekly courses (classroom format)
At events & congresses
Talks and presentations at events and congresses on running your own AI dev team.
Invite me to speakFrequently asked questions
Do I need to know how to code?
Logbook is built for software developers — that's who gets the most out of it. If you write code, you're ready. Not a developer but want to run AI agents on your own work? Book a free intro and we'll see if it's a fit.
What equipment do I need?
Just a computer (Mac, Windows, or Linux) and a stable internet connection. We don't set up everything on day one — we install and connect tools progressively as your needs emerge, session by session.
Is there ongoing support?
Yes. Sessions are pay-as-you-go — book more whenever you need them ($200 each). You keep all the agentic files and configurations we build together, plus access to the Logbook community.
Can I choose my schedule?
Sessions run on weekday evenings, from 5:00 PM (CDMX / Mexico City time) — after your workday. We set your recurring slot in the free intro.
Is the mentorship really 1:1?
Yes. One mentor dedicated exclusively to you, 1:1 — live screen sharing with immediate feedback, throughout your sessions. No group sessions, no recorded classes.
What if I already use AI tools like ChatGPT?
Great — that means you already understand the value. We take you to the next level: from using a web chatbot to having an AI assistant that works directly on your computer, reads your files, executes tasks, and adapts to your workflow through agentic files.
Your mentor is waiting
Schedule a free 30-minute diagnostic session. We analyze your current situation and show you exactly how advanced AI can transform your way of working.
No commitment. No credit card. Just an honest conversation about your productivity.