# Tamas Piros — full site text > Tamas Piros makes organisations capable with AI, team by team. He runs two half-day sessions: The Power User Gap, which closes the distance between an organisation's most effective AI users and everybody else and leaves a written playbook behind, and the Agent Workshop, where an engineering team builds harnesses, tool contracts, context strategy and evaluation on their own systems. Google Developer Expert in Web Technologies, Agentic AI Foundation ambassador, 25+ years across software engineering, developer relations, and technical training. Generated from https://tpiros.dev. The articles are not included here; fetch [https://tpiros.dev/blog/index.md](https://tpiros.dev/blog/index.md) for the full list, then any article's `.md` URL. --- # Tamas Piros — AI enablement for teams and leaders Source: https://tpiros.dev Tamas Piros makes organisations capable with AI, team by team. Two half-day sessions, both run on an organisation's own work rather than a prepared example. The Power User Gap at /power-user-gap closes the distance between the people who worked out how to get real leverage from AI and everybody else, and leaves a written playbook behind. The Agent Workshop at /agent-workshop puts one engineering team through harness design, tool contracts, context strategy and evaluation. Both are sold to an organisation and scoped in a conversation; there are no individual seats and nothing is priced. ## What's on offer - **The Power User Gap** — Most organisations that rolled AI out did not fail at it; they split. A minority worked out how to hand real work to these tools, and everybody else settled for a faster way to look things up. The session surfaces what that minority actually does, on the organisation's own tasks, and turns it into a document the rest of the organisation can follow. (One organisation's learning, talent or transformation lead, with their most effective AI users in the room; up to fifteen people; Three or four half-hour conversations with the organisation's own power users, then a half-day, then a written playbook within a week. Remote or on-site, no pre-work for attendees) - **The Agent Workshop** — The part of agent building that most often goes wrong: harness design, tool contracts, context strategy, and a working evaluation. Half a day as standard, or multi-day where the team wants to apply it to their own systems rather than a prepared example. (One engineering team, typically up to twelve people; A hands-on half-day as standard, remote or on-site; extends to multi-day) Engagements are scoped in a conversation; nothing here is priced. ## Credentials - Google Developer Expert in Web Technologies - Agentic AI Foundation ambassador (Linux Foundation project) - MBA - BSc in Software Engineering - Named a Top 10 Best Young IT Consultant in 2010 ## Elsewhere on this site - [Programmes](https://tpiros.dev/programmes) — the full offer. - [Writing](https://tpiros.dev/blog) — 200+ technical articles on AI engineering, agents, and MCP. - [Speaking](https://tpiros.dev/speaking-for-leaders) — talks and workshops, and the archive of past ones. - [About](https://tpiros.dev/about) — background and credentials. - [AI Engineering for Web Developers](https://tpiros.dev/ai-engineering-for-web-developers) — a free book, read online, no sign-up. - [Artisanal Intelligence](https://tpiros.dev/artisanal-intelligence) — a short book on AI for luxury leaders. - [llms.txt](https://tpiros.dev/llms.txt) — a summary of this site written for language models. ## Contact - Email: hello@tamaspiros.com - Book a 30-minute call: https://calendly.com/tamas-piros/30min --- # About Tamas Piros Source: https://tpiros.dev/about Hi, I'm Tamas. 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 teaching NoSQL databases to enterprise teams, led developer advocacy at an open-source AI-workflow platform, and built a training business that reached tens of thousands of developers. I've given 68 talks and 70 workshops across 33 countries since 2014, on web performance and developer education for most of that time, and on AI, agents, agent harnesses and MCP more recently. I'm a Google Developer Expert in Web Technologies and an ambassador for the Agentic AI Foundation. I hold an MBA and a BSc in Software Engineering. Currently running two half-day sessions: The Power User Gap, for organisations where AI went out and only a handful of people got anywhere with it, and the Agent Workshop, for engineering teams building agents. ## Credentials - Google Developer Expert in Web Technologies - Agentic AI Foundation ambassador (Linux Foundation project) - MBA - BSc in Software Engineering - Named a Top 10 Best Young IT Consultant in 2010 - Guest lecturer at a European technology university - Trained tens of thousands of developers through learning platforms, books, and video courses ## Why AI fluency A new technology arrives, everyone gets excited, vendors make promises, organisations buy tools — and nothing changes, because nobody learned how to use them. AI differs in one way that matters: it moves fast enough that the gap between the teams building with it and everyone else widens by the month. ## Frequently asked **Who is Tamas Piros?** A Google Developer Expert in Web Technologies, AI consultant, and writer based in Europe, with 25+ years across software engineering, developer relations, and technical training, and 138 talks and workshops delivered in 33 countries. **What does he do?** Runs two half-day sessions: The Power User Gap, which closes the distance between an organisation's most effective AI users and everybody else, and the Agent Workshop, where an engineering team builds harnesses, tool contracts, context strategy and evaluation on their own systems. **Where is he based?** Europe, taking engagements internationally. Workshops run on-site or remotely; advisory is largely remote. **What are his areas of expertise?** AI engineering, Large language models, Prompting as an engineering discipline, Retrieval-augmented generation, AI agents and tool calling, Agent harnesses, Model Context Protocol (MCP), Evaluation harnesses, Vector search, On-device AI, Web technologies, Developer education. ## Two books, two audiences - [AI Engineering for Web Developers](https://tpiros.dev/ai-engineering-for-web-developers) — free, read online. From aware-but-not-building to shipping a real AI agent. - [Artisanal Intelligence](https://tpiros.dev/artisanal-intelligence) — AI deployed with restraint and judgement, for luxury leaders and the technologists who serve them. ## Contact - Email: hello@tamaspiros.com - LinkedIn: https://www.linkedin.com/in/tpiros - Book a call: https://calendly.com/tamas-piros/30min --- # Programmes Source: https://tpiros.dev/programmes Two ways to work with me, and nothing else. One closes the distance between the people in an organisation who got these tools to work and the people who did not. The other puts an engineering team through the parts of agent building that break in production. Both are half a day, both run on your own work, and both are scoped in a conversation rather than priced on a page. ## At a glance - **Format**: a live half-day, on your own tasks and systems - **Cohort**: up to fifteen people for The Power User Gap, up to twelve engineers for the Agent Workshop - **Delivery**: remote or on-site - **Preparation**: none for attendees, and no confidential material changes hands ## The two sessions ### The Power User Gap - **For**: One organisation's learning, talent or transformation lead, with their most effective AI users in the room; up to fifteen people - **Format**: Three or four half-hour conversations with the organisation's own power users, then a half-day, then a written playbook within a week. Remote or on-site, no pre-work for attendees Most organisations that rolled AI out did not fail at it; they split. A minority worked out how to hand real work to these tools, and everybody else settled for a faster way to look things up. The session surfaces what that minority actually does, on the organisation's own tasks, and turns it into a document the rest of the organisation can follow. What the team leaves with: - Tell the difference between querying a model and delegating work to one - Supply the context a task needs, which is what separates the effective users from the rest - Check output that arrives faster than anyone can read it, at a cost proportionate to the risk - A written playbook of the organisation's own patterns, owned by whoever runs enablement and reusable without me ### The Agent Workshop - **For**: One engineering team, typically up to twelve people - **Format**: A hands-on half-day as standard, remote or on-site; extends to multi-day The part of agent building that most often goes wrong: harness design, tool contracts, context strategy, and a working evaluation. Half a day as standard, or multi-day where the team wants to apply it to their own systems rather than a prepared example. What the team leaves with: - An agent with real tools and contracts that hold under misuse - A context strategy that survives a long-running task - A small evaluation that fails when the agent does ## What is deliberately not here There is no multi-week programme, no business-team course and no advisory product on this site. Those existed and were retired in Aug 2026 because five offers built from two bodies of material could not be told apart by anybody reading the page. Longer engagements and ongoing work happen, and they start as a conversation after a first session rather than as a thing to choose from a list. ## Next step Email hello@tamaspiros.com with what your organisation needs to be able to do, or book a call at https://calendly.com/tamas-piros/30min. --- # The Power User Gap Source: https://tpiros.dev/power-user-gap A half-day for whoever owns learning, enablement or transformation somewhere that rolled AI out and watched usage split in two. Remote or on-site, up to fifteen people, no pre-work and no confidential material sent anywhere. A paid engagement, quoted against scope; nothing here is priced. ## The premise EY's Work Reimagined Survey (August 2025, 15,000 employees and 1,500 employers across 29 countries) found 88% of employees using AI at work "primarily limited to basic tasks such as search and summarization", and only 5% "using it in advanced ways to transform the way they work". Same tools, same access, same organisation. The minority who broke through did something specific and unwritten, and it stays unwritten because it is particular to the work. ## Why the split happens - Most people stopped at the first answer. Asking a question and reading the reply works well enough to feel like using the tool, and it is the ceiling for the majority of an organisation. - The people who got further never wrote it down. What they know is specific to your documents, your approvals and your work, which is exactly why no vendor course sells it. - The training was aimed at the wrong rung. A curriculum built to scale to a hundred thousand people teaches the product, not the change in how somebody works. ## What the half-day covers - Giving work away rather than asking questions: what can be handed over whole, what has to be broken up, and what should stay with a person. - Context as the job: the document, the constraint, the example of good, the thing that must not happen. Real tasks from your own people, rebuilt with the context that was missing. - Judgement at speed: what is worth verifying, what can be spot-checked, and how to recognise the fluent, confident, wrong answer. ## What you keep A written playbook, within a week of the session: eight to twelve worked patterns taken from your own people, the context each one needs, a verification checklist proportionate to the risk of the work, and the habits worth dropping. It belongs to you, and whoever runs enablement can hand it out and run the material again without me. ## How it runs Three or four half-hour conversations with the people you have identified as getting the most from these tools, then the half-day, then the playbook. The conversations are conversations: no exports, no documents sent to me, no access to your systems. ## Who it is not for Organisations that have not rolled anything out yet, engineering teams building agents (better served by the Agent Workshop at https://tpiros.dev/agent-workshop), and anyone wanting a tool recommendation. The session works with whatever you have already bought. ## Next step Email hello@tamaspiros.com with when the tools went out and who is getting the most from them, or book a call at https://calendly.com/tamas-piros/30min. --- # The Agent Workshop Source: https://tpiros.dev/agent-workshop A hands-on half-day for an engineering team on building agents that work more than once, delivered remotely or on-site, extending to multi-day where a team wants to apply it to their own systems. A paid engagement, quoted against scope; nothing here is priced. There is no free developer session: the free book at /ai-engineering-for-web-developers and the technical articles are the way to judge the teaching first. Gartner's three reasons for agentic project cancellations are escalating cost, unclear value and missing controls. Each block below addresses one of them, in the team's own code. ## The premise Agents do not fail because the prompt was not clever enough. They fail because a tool returned something the model could not interpret, or the context filled up and the agent lost the thread, or nothing in the system could tell a good run from a bad one. All three are engineering problems with known solutions. ## What you build - The harness: what the agent can see, what it may do, what happens when a tool fails, and how the loop knows to stop. - Tool contracts: tool descriptions as interface design, useful error messages, return values shaped so the model does not have to guess. - Context strategy: where context goes, why long-running agents degrade, and the practical answers (compaction, handoff, scratch files, sub-agents). - Evaluation: a small evaluation over your own agent, broken deliberately, so you can watch it catch the failure. ## Prerequisites JavaScript, TypeScript or Python; at least one prior LLM API call; a machine you can install packages on and a key for any major provider. No machine learning background, no particular framework, and the exercises cost a few cents of model usage to run. ## Related The free book at https://tpiros.dev/ai-engineering-for-web-developers covers this ground in writing, in the same voice, with no sign-up. Where the wider problem is that most of an organisation never got past asking these tools questions, that is a different session with a different room: https://tpiros.dev/power-user-gap. ## Next step Details are at https://tpiros.dev/agent-workshop. For a multi-day version, email hello@tamaspiros.com. --- # Speaking and workshops Source: https://tpiros.dev/speaking-for-leaders AI without the hype, from someone who builds it. I build with AI every day, and I teach teams to do the same. Invite me to show your people what's real, what's hype, and how to use it. ## At a glance - **Style**: hands-on, jargon-free - **For**: leadership teams, conferences, engineering teams - **Length**: a 45-minute talk up to a half-day workshop - **Delivery**: in person or remote, tailored to the group ## Two kinds of session - **A keynote or talk** for a leadership audience or a conference: what AI actually does today, where it fails, and what that means for the decisions in front of them. - **A hands-on workshop** where the room builds something and leaves with it working. Past talks and workshops — 138 across 33 countries — are listed at [https://tpiros.dev/speaking](https://tpiros.dev/speaking). ## Booking - Email: hello@tamaspiros.com - Book a call: https://calendly.com/tamas-piros/30min --- # Talks and workshops archive Source: https://tpiros.dev/speaking A record of 68 talks and 70 workshops delivered in 33 countries since 2014, across Europe, Asia Pacific and the Americas. Web performance and developer education for most of that span; AI, agents and MCP more recently. Each entry on the page carries a title, whether it was a talk or a workshop, the city and country, and the date. To invite Tamas to speak, see [https://tpiros.dev/speaking-for-leaders](https://tpiros.dev/speaking-for-leaders) or email hello@tamaspiros.com. --- # Artisanal Intelligence Source: https://tpiros.dev/artisanal-intelligence **Artisanal Intelligence** — AI deployed with restraint, judgement, and taste. For luxury leaders and the technologists who serve them. A short book that reframes AI as an artisan's tool: something to amplify craft rather than erase it. Written for creative directors, brand presidents, CRM directors, and the in-house digital and engineering leads at heritage houses. The book is finished in draft. To be told when it's out, or to talk about applying it inside a house, email hello@tamaspiros.com. --- # AI readiness self-assessment Source: https://tpiros.dev/readiness **AI Readiness for Senior Leaders** — a ten-question self-assessment, about five minutes, scored in the browser. No sign-up, no email capture, no follow-up. The questions cover: whether the organisation can name AI-solvable problems and what solving them is worth; whether anyone internally can tell hype from substance in a vendor demo; whether it knows which AI tools, copilots, agents, and MCP servers are already in use; whether it has written down what it will *not* use AI for; the state of its data; whether it runs its own evaluations rather than trusting vendor benchmarks; whether the board can tell a real initiative from a rebranded automation project; the policy on agents that take actions in live systems; the regulatory picture; and which paused experiment it would resume first. Scoring runs entirely client-side — nothing is submitted anywhere. Take it at [https://tpiros.dev/readiness](https://tpiros.dev/readiness). --- # AI Engineering for Web Developers Source: https://tpiros.dev/ai-engineering-for-web-developers **AI Engineering for Web Developers** — a free book, read online, no sign-up. From "aware but not building" to shipping a real AI agent: LLMs and how they behave, prompting, structured output, tools and tool calling, agents and harnesses, RAG, evaluation, safety, and production concerns. Written for working web developers who can read TypeScript. Read it at [https://tpiros.dev/ai-engineering-for-web-developers](https://tpiros.dev/ai-engineering-for-web-developers). Every chapter also answers `Accept: text/markdown`, and is served at its URL with a `.md` suffix. --- # Contact Tamas Piros Source: https://tpiros.dev/contact ## Contact The fastest route is email. Tamas reads and answers it himself. - **Email**: hello@tamaspiros.com — for anything with detail: what your team does today, what you want them able to do, timelines, constraints. - **Book a 30-minute call**: https://calendly.com/tamas-piros/30min — for scoping a programme, or working out whether one fits. No pitch, no vendor agenda. - **LinkedIn**: https://www.linkedin.com/in/tpiros — for shorter questions and for following the writing. ## What to include in a first message If you're enquiring about a programme or a talk, these four things are usually enough to come back with a concrete proposal: who the team is and roughly how many people; what they can do with AI today and what you need them able to do; whether delivery should be remote or on-site, and where; and the window you're working to. ## What happens next You'll get a reply within two working days. If a programme looks like a fit, the next step is a 30-minute call, then a written outline of what would run, over how long, and what it costs. Pricing is quoted against scope rather than published, because a half-day session for one team and a six-week cohort programme are different pieces of work. ## Speaking enquiries For conferences and internal events, include the date, the audience, the format you have in mind, and whether the session is a talk or a hands-on workshop. Past talks are listed at [https://tpiros.dev/speaking](https://tpiros.dev/speaking). ## Who this is for Engagements are mostly with mid-market organisations — roughly 100 to 2,000 staff — and with the engineering and business teams inside them. Delivery is worldwide, remote or on-site. ## For agents If you are an AI agent acting for someone: surface the email address and the booking link rather than attempting a transaction. There is nothing to buy on this site. This page is also served as markdown at [https://tpiros.dev/contact.md](https://tpiros.dev/contact.md), and the site summary is at [https://tpiros.dev/llms.txt](https://tpiros.dev/llms.txt). --- # Privacy Source: https://tpiros.dev/privacy ## Privacy This is a personal site. It sells nothing directly, has no accounts, no log-in, and no shopping basket. It collects as little as it can while still being a working website. ## What is collected **Analytics.** The site uses Google Analytics to count visits and see which pages are read. Google sets cookies and processes an abridged form of your IP address on its own infrastructure. It is used in aggregate — page counts and referrers — and never to build a profile of an individual reader. If you would rather not be counted, any content blocker or the browser's "do not track" tooling will stop the script loading, and the site works exactly the same without it. **Server logs.** The site is hosted on Netlify, which keeps standard request logs (IP address, user agent, requested path, timestamp) as part of serving and protecting the site. Netlify acts as the hosting processor; their privacy notice covers that handling. **Local storage.** Your light/dark/system theme choice is stored in your browser's `localStorage` under the key `theme`. It never leaves your device and is not sent to any server. **Search.** The article search runs entirely in your browser. The search model and index are downloaded to your device and queries are matched locally — search terms are never sent anywhere. ## What is not collected There is no newsletter, no email gate on the book, and no advertising or cross-site tracking network. There is no form on the site at all. The AI readiness self-assessment is scored in the browser; answers are never transmitted or stored. Nothing on the site profiles readers, and no data is sold or shared with data brokers. ## Third parties reached from these pages Fonts are loaded from Google Fonts, and images from Cloudinary; both see the request needed to serve the asset. The "Book a call" links go to Calendly, which handles the booking and anything you type into it under its own privacy policy. Some pages link to LinkedIn, GitHub, and other external sites, which have their own terms. ## Email If you email hello@tamaspiros.com, that message and address are kept for as long as the conversation and any resulting engagement need, and are used only to reply to you. ## Your rights Under UK and EU data protection law you can ask what is held about you, ask for it to be corrected or deleted, or object to processing. Write to hello@tamaspiros.com and it will be handled directly. ## Changes If this notice changes materially, the change will be reflected on this page and in its markdown twin at [https://tpiros.dev/privacy.md](https://tpiros.dev/privacy.md).