When IKEA automated 8,500 customer-service roles, it did not lay those people off. It reskilled them as interior design advisors — and they went on to generate roughly $1.4B in new revenue.
That is the choice in front of every technology leader right now. Not whether AI changes your team's work — it already has — but whether your existing people move up the value chain or get routed around.
This training exists for one outcome: your team, up the value chain.
>80%
of AI projects fail — twice the failure rate of non-AI IT projects.
~30%
of GenAI projects were forecast to be abandoned after proof of concept by the end of 2025.
The five in a hundred that report P&L impact.
95%
of enterprise GenAI pilots show no P&L impact, self-reported.
The gap is not tools — your team has the tools. The gap is skills-in-context: nobody showed your developers how to use AI on your stack, your conventions, your development cycle. Training in context is the cheapest fix to the most expensive problem.
Half or full day
What your team does on Monday changes.
2 to 4 weeks
A real use case shipped in your environment.
2 to 4 weeks
New staff never needs a vendor invoice.
Every engagement is fixed-price and scoped in writing.
The full-day agenda. Every block ends with the participants doing the work themselves, on the exercises and datasets they take home.
What vibe coding is and why leading engineering organizations adopted it; the enterprise AI platform landscape (Claude, Azure AI Foundry and Copilot, Gemini, Bedrock, OpenAI); managing outsourced development with AI; when not to use AI — security, compliance, sensitive data; open discussion of the fears and objections in the room.
Live demos generating, optimizing and debugging SQL queries and DAX measures from natural language. Each participant then solves a real scripting problem on the synthetic dataset with a step-by-step guide.
What a skill is and the anatomy of a good one; a live demo building a SQL/DAX skill to team standards; each participant builds a skill for their own daily work from ready-made examples (sql-review, doc-sp, test-generator, spec-writer).
Prompt vs skill vs agent; agent architecture — tools, memory, planning, feedback loops; a live demo of an agent that writes, executes, validates and fixes SQL; each participant designs an agent on a design sheet, from ready-made examples (revisor-codigo, generador-tests, a qa-entregas orchestrator).
Break.
Requirements gathering, test automation, legacy documentation, AIOps, vendor-delivery review and automated code review. In the guided lab each participant picks a use case and starts an MVP, which continues as a two-week assignment and is reviewed in the follow-up Q&A.
Managing outsourced development with AI; which work to insource first; a 30-60-90-day adoption plan the sponsor can act on; final Q&A.
This is not a tool demo. Every participant leaves having built a skill for their daily work and designed an agent for a real use case — and the session ends with a 30-60-90 adoption roadmap your CIO can act on, including the conversation vendors skip: when not to use AI.
Miguel Fierro spent ~10 years at Microsoft working on AI as a Forward Deployed Engineering Manager — participating in over 100 projects deploying AI workloads inside customer environments and generating over $500M of business impact. He is the creator of Recommenders, the most popular open-source recommendation library on GitHub. He has made over 200 interviews for AI profiles. He holds a PhD in Robotics (UC3M with King’s College London, best doctoral thesis award) and executive education from MIT Sloan. He delivers personally.
The same person who delivers Forward Deployed AI Engineering engagements teaches this: the exercises come from production work, not from a curriculum vendor.
Miguel Fierro on LinkedInThe current workshop template has been built and delivered end-to-end for the technology team of an enterprise client: live hands-on exercises on a synthetic banking dataset, ready-made skills and agents, a facilitator script, and a follow-up session.
Vendor training sells you the vendor platform: it is a tool demo, not a change in behavior. This workshop is platform-agnostic, runs on your own stack and conventions, and includes the when-not-to-use-AI compliance block that vendor sessions skip.
Self-paced courses have single-digit completion and zero context. Nobody builds a skill for your SQL conventions, or an agent for your vendor-delivery review, from a MOOC. This is live, on your use cases, with the people who own the roadmap in the room.
Tool overviews and development-cycle integration are different products. If the previous workshop had worked, your developers would be using AI in the development cycle today. This workshop is built around the cycle itself — your stack, your conventions, your use cases — and ends with a 30-60-90 roadmap.
The exercises run on a synthetic dataset, an NDA is standard, everything built in the session belongs to you while the IP in the training materials stays with the instructor, and the instructor keeps no access to your systems afterward.
Price it against one failed AI pilot — the failure rates above — or against one month of one outsourced developer. The 30-60-90 roadmap block alone changes what the decision is being compared against.
