Tamas Piros
I make organisations capable with AI, team by team.
Most rollouts do not fail, they split: a handful of people work out how to hand real work to these tools, and everybody else settles for a faster way to look things up. Two sessions close that gap, one for an organisation and one for an engineering team, both run on your own work.
I build with AI every day. 25+ years building and teaching technology. Google Developer Expert and Agentic AI Foundation ambassador. I've taught tens of thousands of developers and delivered 138 talks and workshops in 33 countries.
Programmes
Two sessions, on your own work
One for an organisation where the tools went out and only a handful of people got anywhere with them. One for an engineering team whose agent works on a good day. Both run on your tasks and your systems rather than a prepared example.
For organisations whose AI rollout split in two
The Power User Gap
88% of employees use AI at work and 5% use it in ways that change the work. A session with the people who worked it out, on their real tasks, and a written playbook of what they do for everybody who has not. Remote or on-site, nothing to prepare.
What the session covers →For an engineering team
The Agent Workshop
Harness design, tool contracts that hold when an agent misuses them, a context strategy that survives a long task, and an evaluation that fails when the agent does. Your engineers build, break and fix all four on their own stack.
What the session covers →How it fits together
Agentify your organisation
The two sessions are where most engagements start. The work around them is agentifying how the place actually runs, one person, one team, or the whole organisation.
One person
Power users, made on purpose
One person, their real week, and the tasks in it that an agent should be doing by Friday. The same method as the Power User Gap, applied to a desk rather than a room.
One team
A workflow with agents in it
Take one thing a team does every week, the ticket triage, the report, the onboarding pack, and rebuild it so an agent runs the parts that don't need a person. The team ends up with a workflow it owns and can change, rather than a demo somebody else built.
The organisation
Transformation that survives Monday
AI transformation is a change management problem before it is a technology one. Who goes first, what gets measured, what you tell the people who are worried, and the playbook that carries it. I work alongside whoever owns the change, from the first pilot to the point where nobody calls it a pilot any more.
None of this is priced on a page. It starts as a conversation, usually after a first session.
Tell me what you are trying to change →About
How I got here
I've spent 25+ years across software engineering, developer relations, and technical training, working across Europe, Asia Pacific and the Americas. Most recently I led regional developer advocacy at a cloud media platform, running AI workshops and enablement programmes for engineering teams across the region. Before that I was a principal technical instructor, taught NoSQL databases to enterprise teams, and built a training business that reached tens of thousands of developers.
I'm a Google Developer Expert in Web Technologies and an Agentic AI Foundation ambassador, with an MBA and a BSc in Software Engineering. I've published technical books and video courses, given guest lectures at a technology university, and have delivered 138 talks and workshops in 33 countries.
I build with AI every day: LLMs, vector search, agentic systems, MCP servers, on-device inference. I use agents in my own work daily, and I help organisations do the same, from one person's week to a whole rollout and the change management around it. Everything I teach comes out of that daily work.
Writing
Selected writing
Selected pieces on agents, MCP, and building real AI systems. More in the full archive.
The forward deployed engineer land grab
Four companies committed roughly $9 billion in nine weeks to putting their own engineers inside customer organisations. What a forward deployed engineer actually costs, who really gets one, and the question to ask before you sign anything: what remains when they leave?
What is agentic AI? A working definition for 2026
A direct definition of agentic AI, what's actually shipping in production today, what isn't, and the five things worth building first. With specifics, citations, and an honest list of what doesn't work yet.
Harness - A Local Managed Agent in Four Planes
Anthropic decoupled their agent into brain, hands, and session. I built a local TypeScript version in roughly 2,400 lines to see what the architecture actually feels like when you type it out.
MCP Servers - The Bridge Between LLMs and Real-World Tools
Model Context Protocol (MCP) servers expose tools, resources, and prompts to LLMs in a unified, structured way. This post explores how they work, how to build one, and why they are a critical part of the future AI stack.
Filesystem as Context: Building an AI Detective with bash-tool
Instead of stuffing documents into prompts, give your AI agent a filesystem and let it retrieve its own context. Here's how, using a murder mystery detective as the demo.
Talks
Speaking
I've spoken at technology conferences worldwide on AI, agents, agent harnesses, MCP, web technologies, and developer education. Recent topics include building with LLMs, agentic systems, and how AI fits into engineering and creative practice.
138 engagements across 11 years
- 20268
- 202517
- 202414
- 202316
- 202214
- 202128
- 202021
- 201913
- 20162
- 20153
- 20142
- Workshop
AI Crisis
Coventry, UK
- Workshop
Demystifying AI Agents: A Hands-On Workshop
Frankfurt am Main, Germany
- Workshop
Build a Production-Ready AI Agent in 90 Minutes
Berlin, Germany
- Workshop
Build a Production-Ready AI Agent in 90 Minutes
Berlin, Germany
- Workshop
Building Real AI Agents with Gemini
Pristina, Kosovo
- Workshop
The name is Agent. AI Agent.
Rome, Italy
- Workshop
The name is Agent. AI Agent.
Skopje, North Macedonia
- Workshop
The name is Agent. AI Agent.
Warsaw, Poland
- Workshop
Practical AI for Web Developers
George Town, Malaysia
- Workshop
The MCP Workshop: Turning APIs into AI-Ready Tools
Taipei, Taiwan
Books
Two books, two audiences
I'm writing for the two groups I spend most of my time with: the engineers building AI, and the luxury houses deciding how to use it.
For engineers · Free, read online
AI Engineering for Web Developers
From aware-but-not-building to shipping a real AI agent. A hands-on path through LLMs, prompting, tools, agents, RAG, evals, and production, written for working web developers. No sign-up.
Start reading →For luxury leaders · Final draft
Artisanal Intelligence
AI deployed with restraint and judgement, so it amplifies craft rather than erasing it. For luxury leaders and the technologists who serve them.
Read the thesis →Contact
Tell me which room you have in mind, a session for your organisation, a workshop for your engineers, or a talk, and I will come back with a plan.