Leading technical decisions in the AI era
Code is cheaper than ever. Bad technical decisions are not.
Your team is producing more than ever. Reviews pile up, knowledge concentrates, and no one is sure if AI is genuinely helping or only creating an illusion of speed. This workshop gives your team the tools to regain control.
Delivered in Spanish by default. English delivery may be available depending on context.
Talk about my team →The workshop at a glance
- Who it's for
- Engineering teams running AI in production
- Format
- Full day or two half-days
- Size
- 8 to 20 participants
- Mode
- Remote · on-site on request
- You take
- PR review checklist + Decision Canvas
I've shared these ideas at
The problem isn't producing more code
AI lets teams generate more code, faster, with less friction. Understanding, reviewing, operating, and maintaining that software is still human, contextual work.
The risk isn't only writing bad code. It's deciding more things implicitly, with less shared context, and leaving invisible debt that surfaces three months later in an incident.
- Your PRs get approved because they compile, not because anyone understood the decision hidden inside.
- Your reviews stay on how it's written, not on what risk you take when merging.
- Your critical knowledge lives in three heads and evaporates when one rotates out.
- Your architecture drifts from small omissions no one caught in review.
- Your seniors find out too late about decisions that should have been discussed up front.
For teams already feeling the strain
This workshop fits if your team is already using AI (or starting to) and you want to avoid speed turning into loss of judgment, ownership, or maintainability.
- CTO or Head of Engineering: you want to define internal practices for responsible AI use before the team improvises five different approaches.
- Engineering Manager: you need to align review and decision criteria across the team.
- Staff Engineer or Tech Lead: you want to turn your individual judgment into shared team practice.
- Your team is scaling product, architecture, or contributor count. You notice you no longer hear about every decision being made.
What we work on in the workshop
Which decisions are worth a conversation before you merge
How to tell trivial decisions apart from those worth attention before they turn into expensive debt. Reversibility, cost of change, operational risk, uncertainty, context.
How to review what AI has already written
How to review intent, context, verifiability, operational impact, and ownership in AI-assisted changes. Beyond style and implementation.
Debt isn't born from one bad decision; it grows from many small omissions
How to detect debt born not from one big bad decision, but from many small omissions no one discussed in time.
From individual judgment to team practice
How to turn senior judgment into agreements, checklists, and conversations the whole team can reuse without relying on heroics.
What the team takes away
You don't leave with loose ideas. You leave with two practical artifacts built on your own cases.
AI-assisted PR Review Checklist
A practical checklist for reviewing AI-assisted changes beyond style or implementation: intent, context, risk, verifiability, ownership.
You work it through during the session on your own PRs. You take the template with you as a Google Doc.
Technical Decision Canvas
A lightweight canvas for discussing technical decisions across reversibility, cost of change, operational risk, uncertainty, and context.
You build it during the session on the team's real decisions. You take the canvas with you as a Google Doc.
The point isn't adding bureaucracy. It's knowing which conversations are worth having before speed turns into debt.
Format
Private workshop for the company, adapted to the team's context.
- Full day or two half-days.
- Remote by default. On-site possible depending on availability.
- Recommended for 8 to 20 participants.
- Light preparation with the team beforehand.
- 80% proven structure, 20% adapted to your context.
Where context calls for it, a follow-up session afterwards can be added.
Who runs it
Emilio Carrión
Staff Engineer · Mercadona Tech
I support 7 engineering teams in Mercadona Tech's online division: logistics, fulfillment, and warehousing. Systems that process tens of thousands of orders every day for Spain's largest retailer. I don't talk about architecture in the abstract — I talk about the architecture that moves lettuce. I'm also a PhD candidate at Universitat Politècnica de València researching software production methods, and I write weekly for thousands of senior engineers about architecture, technical debt, technical judgment, and engineering in the AI era.
This workshop comes from a practical obsession: how teams that build software meant to survive production make good decisions.
What it isn't
- Not a prompt-engineering course.
- Not a tooling demo.
- Not a futuristic talk about the end of programmers.
- Not generic productivity training.
- Not abstract architecture consulting.
It's practical work on technical judgment, review, verification, debt, and decision-making in real teams.
Frequently asked
▸Does the team need prior experience with AI?
The workshop is designed for teams already using AI day to day. It is not an intro to tools — content assumes you use them and focuses on judgment, review, and decisions.
▸Can it be adapted to our stack and context?
Yes. Before the session I review your context and adapt exercises and examples. The structure is stable; the cases are yours.
▸Do you issue invoices?
Yes. Invoice with VAT breakdown for Spanish companies and intra-community invoicing for the rest of the EU.
▸How do I justify it internally?
A senior easily spends five hours a week on reactive work (reviews without context, repeated knowledge, decisions without a system). In a team of five, that is roughly twenty-five hours a week recovered. The workshop pays back in less than a month.
Does this fit your team?
If you're figuring out how to use AI without losing technical control, or if your team needs to improve how it reviews, decides, and shares context, drop me a line and we'll see if it fits.
Talk about my team →