Four AI workshops for your team
Four workshops for different groups of people: management, admin and marketing, business teams, and developers. Every one of them is hands-on, runs at your office, and does not end with the last slide. Four weeks later we come back for a follow-up.
Workshops and AI projects delivered for, among others:
Which workshop is for you
Each workshop is aimed at different people. If you are not sure which fits, we go through it on a short call.
| Workshop | Who for | Length | Format |
|---|---|---|---|
| How modern teams use AI | Management and department heads | 2–3 hours | At your office or online |
| Vibe coding in business | Managers and senior business people | 1–2 days | In person at your office |
| Agentic engineering | Developers and product teams | 1–2 days | In person at your office |
| From everyday AI chat to advanced skills and agents | Admin, marketing and sales | 4–8 hours | In person at your office or online |
Prices are further down this page.
Workshop 01
How modern teams use AI
We walk you through nineteen AI rollouts we built over the past year, in companies ranging from an estate agency through hospitality to HFT trading. For each one you hear what was squeezing the company, how we solved it, and what to take away. You leave with a clear sense of where AI belongs in your company and where it does not yet.
- Length
- 2–3 hours
- Format
- At your office or online
- Who for
- Management and department heads
- Price
- from €1,500
What we cover
- The L0 to L4 maturity ladder: from plain chat through custom GPTs and automated workflows to processes built around AI. You get the vocabulary that lets your team agree on where you actually stand.
- An exercise where everyone places their own department on the L1 to L4 scale, followed by a discussion of what can be improved and how to move up a level.
- Nineteen projects, each as situation, solution and lesson. An estate agency with fifteen agents whose commercial CRM ate €15,000 a year and still did not do what they needed. A controlling team retyping reports from fifteen sites by hand. A restaurant where the manager could not keep up with enquiries.
- Twenty website designs for a thousand dollars in tokens instead of an agency tender. A marketing GPT that learned the founders’ tone from transcripts of their own conversations. A personal assistant on an in-house server that watches both calendar and inbox.
- A good prompt and a bad one side by side, using a customer complaint as the example. And the question that moves a company furthest: how do we give AI enough context to do this on its own from now on?
- Who ends up taking the saved time: the employee, the supplier or the company? If ten hours become two, the other eight should be your decision, not an accident.
- Why you can only save once while growth compounds. Instead of “we have four accountants, let us get to two”, we ask what you would want ten times more of.
What the team leaves with
- You will know where each of your departments sits on the L0 to L4 scale, and which one is furthest from what is possible today.
- A list of processes AI can improve now, ordered by what is worth starting with.
- A sober sense of cost and time: what you can do in an evening, what is a month of work, and what needs your own people.
- A shared language, so AI can be discussed in the company without unnecessary confusion.
- Clarity on what you can handle yourselves and when it is worth calling someone in.
How it runs in the room
- It is not a monologue. Most of the questions come up during the individual projects, so we leave plenty of time for them.
- We show real screens from applications and agents running at clients. No stock illustrations.
- One short exercise where you place your own departments. That is usually where the best discussion starts.
Where we roll this out most often
The departments with the most repeated manual work, where saved time shows up first.
Operations and processes
For teams that move data between different tools every day and handle routine admin.
What they learn: How to connect your systems into a single flow. AI prepares the inputs and sorts the information, so all that is left for you is to approve it.
Sales and marketing
For people who spend hours sorting enquiries, writing proposals or producing content.
What they learn: How to have enquiries sorted automatically, proposal drafts prepared, and communication tailored to each client.
HR and recruiting
For recruiters and HR managers who need to cut the time from an incoming CV to the first interview.
What they learn: How to let AI summarise dozens of CVs and prepare a clear shortlist of candidates.
Management and reporting
For managers and analysts who lose time stitching spreadsheets together by hand.
What they learn: How to have recurring management reports assembled automatically, with no transcription errors.
Michal Juhás AI consultant Workshop 02
Vibe coding in business
A full day at the keyboard. You go through four rounds of building and leave each one with something that actually runs. In the morning you tackle your own routine, in the afternoon what the customer feels.
- Length
- 1–2 days
- Format
- In person at your office
- Who for
- Managers and senior business people
- Price
- from €3,000
What we cover
- Two ways to use AI. Either you build software with it that then runs on its own and always the same way, or you build an application that uses AI at run time. You see the difference across three things: a reporting dashboard, an analytics tool and an autonomous agent.
- The morning is about your own productivity: getting on top of the inbox, management reporting, preparing for meetings, working with spreadsheets, a research assistant, or searching internal documents.
- In the afternoon you build what the customer feels: personalised recommendations, smarter support, a campaign generator, or tools built on what you already know about your customers.
- Why access to data helps an agent most. How it works with inboxes, databases, documents and APIs, and how structured data differs from unstructured. And why more data does not automatically mean a better result.
- Three ways to build: vibe coding, spec-driven development and agentic engineering. When each one fits, and how to tell where your idea belongs.
- Three places it can run: on a laptop, as a managed agent in the cloud, or as a hosted application. For each we cover the upsides, the costs, and what you have to watch.
What the team leaves with
- Working prototypes your people built with their own hands during the day.
- Shared repositories with CLAUDE.md, a ready vibe-coding skill, security rules and a testing checklist. In other words: a method the team can repeat without us.
- A view of which prototypes are worth finishing, which can stay small helpers, and which you can drop.
- A sober estimate of what you can solve this way in an hour and what is weeks of work.
How it runs in the room
- You work in groups of three or four. First you go through the task and a possible solution together, then you build alone or in pairs. The more experienced get a harder variant so they are not bored.
- A short explanation right before you start, then up to 90 minutes of building. After each round you show each other what you made and we take it apart together: what came out, where it creaked, and whether it is usable.
- We move between the groups, help with technical snags and ask the uncomfortable questions. For larger teams both of us come.
- Nobody starts from a blank page. Everyone begins from a prepared repository with sample data: inboxes, spreadsheets and product records.
- We work on invented or pre-approved data. No company or customer records without your consent, and no access keys in the code.
What to prepare
- Two to three hours of preparation are included. We meet, go through what slows the team down most, and choose the tasks accordingly. If you want the workshop built entirely on your own processes and data, we prepare it in depth as an add-on.
- Every participant needs a laptop, an active Claude subscription, Claude Code installed, and a GitHub or GitLab account.
- Three days ahead, everyone runs a short test: clone the repository, start Claude Code, make a small change and push it to GitHub. One exercise confirms that everything works for everyone.
- If git is new to part of the team, we add a short preparation session before the workshop.
Michal Juhás AI consultant 
Josef Nevoral CTO, agentic engineering Workshop 03
Agentic engineering
For teams that already use AI but still write the code by hand. The goal is plain: your developers write not one line manually. Instead they build a system that produces good software.
- Length
- 1–2 days
- Format
- In person at your office
- Who for
- Developers and product teams
- Price
- from €3,000
What we cover
- Four questions that lead from prompting to engineering. How do you ask the model? What does it need to know? How do you describe precisely what should be built? And how do you design an environment where agents can be trusted?
- Context engineering: what the agent needs to know about your code so it does not miss. Specifications, tickets, logs, examples of what a good result looks like, and user feedback.
- Harness engineering: the guardrails inside which an agent works on its own and still safely. Tests, code review, CI/CD, monitoring and security checks.
- Spec-driven development in practice. How an idea becomes a brief, user requirements, acceptance criteria, a technical specification and tests. We show it on BMAD and GitHub Spec Kit.
- What breaks when companies rush this, and how you avoid it.
- A section for management: why the man-day stops making sense as a billing unit once agents run in the background, and what a team’s token consumption reveals about how fast it is learning.
What the team leaves with
- A configured environment where agents run directly on your code.
- The rules and context the agent relies on, written down so the team can keep adding to them.
- An agreement in the team on what you let agents do and where a human still decides.
- A view of the costs: what tokens run to per developer per month, and what the company gets for it.
How it runs in the room
- A short explanation, then work happens on your code, not a practice project. After each round we look together at what the agent produced and whether it would pass your review.
- Two of us run it. Michal takes processes and costs, Josef architecture and engineering guardrails. For larger teams both of us come.
- We work on invented or pre-approved data. No access keys in the code and no production records without your consent.
What to prepare
- A laptop with a development environment, an active Claude subscription, Claude Code installed and signed in, and access to the repository we will be working in.
- We go through your code and processes beforehand, so the tasks match what you are actually building.
Hear it from us
Michal and Josef discuss exactly what this workshop covers: how a developer’s work changes when agents write the code.
Josef and Michal ran a hackathon for our engineers. Going in, they knew very little about agentic engineering. By the end of the day they had AI autonomously building a new feature, from the ticket all the way to a merged PR.
Martin Valovsky Defence Tech & Brain Health Entrepreneur, SEC Technologies Michal Juhás AI consultant Josef Nevoral CTO, agentic engineering Workshop 04
From everyday AI chat to advanced skills and agents
Most teams use AI as a better Google. This workshop moves them from random chatting to systematic work: better prompts, a shared knowledge base, and the first automations.
- Length
- 4–8 hours
- Format
- In person at your office or online
- Who for
- Admin, marketing and sales
- Price
- €1,500 – €3,000
What we cover
- How ordinary chatting differs from systematic prompting. Why two people with the same tool get completely different results.
- How to get from chatting to automations, meaning things that then run on their own.
- How to prompt substantially better. Not by a few percent, but so the difference is obvious at a glance.
- A knowledge base for the team: what belongs in it and how it is maintained so people actually use it.
- A shared repository of skills and prompts, so good practice does not stay in one person’s head.
- Practical challenges where everyone tries several approaches and compares which one worked.
What the team leaves with
- Concrete ideas for what to do differently next time you reach for AI.
- A framework for choosing the right approach for a given situation.
- Ready-made prompts for marketing, sales and operations.
- A list of tools and clarity on what each one is good for.
How it runs in the room
- A short presentation to open, then straight into the work.
- A list of challenges, each with a fixed time.
- Participants work through the challenges, and each one is followed by a short debrief with the instructor.
What to prepare
- A ChatGPT, Gemini or Claude subscription for every participant.
- We go through your processes and tools beforehand, so the tasks match what you actually work with.
Michal Juhás AI consultant or 
Andrej Winter Consultant, data and AI How it works
Four steps, no surprises
- 01
A 15-minute call
We work out which workshop makes sense for you. If none of them does, we say so plainly.
- 02
Short preparation
You send us some material and context. We prepare exercises built on your processes and your terminology.
- 03
The workshop at your office
No slides about the future. The whole time we work on your cases and your data.
- 04
Follow-up after 4 weeks
A short meeting on how it is going in practice. We sort out whatever got stuck in the meantime.
What you are comparing this with
Before you look at the prices below, here are the other routes companies consider, and what they cost.
Nothing
We are waiting for it to mature
- The team uses ChatGPT on personal accounts and nobody has any visibility.
- Your competitors pull a little further ahead every month.
- No measurable result.
YouTube / do it yourself
We will hand it to a bright junior
- Cheap.
- Hours spent working out which parts are even true.
- No priorities. The team starts twenty things and finishes none.
Big 4
Deloitte, PwC, EY, KPMG and the rest
- A brand your board recognises.
- €30–80k, six months, and a team of juniors on the project.
- At the end you have an 80-page strategy. Rolling it out is on you.
AI Training for Teams
Run by people who work with AI every day
- We work with AI daily and build our software with agents.
- AI rollouts tailored directly to your processes, no 80-page strategy.
- A follow-up after four weeks, so things actually get finished.
Workshop prices
The price is one-off and covers the whole team, not each person. The more colleagues come, the cheaper it works out per head. We agree the final figure before the workshop and it does not move; add-ons are priced separately.
How modern teams use AI
2–3 hours · Management and department heads
- Two to three hours of preparation, going through your processes
- An exercise where everyone places their own department on the L0 to L4 scale
- Nineteen real rollouts from the past year, including the ones that did not work
- Materials, prompts and examples in English, yours to keep
- Led by Michal Juhás, not a junior
- Follow-up meeting four weeks later
Vibe coding in business
1–2 days · Managers and senior business people
- Two to three hours of preparation: we go through your processes and pick the tasks
- A technical dry run three days before, so we do not spend the day sorting out access
- Tasks from your own company, not textbook examples
- Working prototypes your people build with their own hands
- Shared repositories with CLAUDE.md, prompts and security rules
- Led by Michal Juhás and Josef Nevoral
- Follow-up meeting four weeks later
Agentic engineering
1–2 days · Developers and product teams
- Two to three hours of preparation, going through your code and processes with you
- A technical dry run three days before, so the environment works for everyone
- We work in your repository, not a practice project
- Guardrails in place: tests, code review, CI/CD and monitoring
- Shared repositories with CLAUDE.md, skills and a testing checklist
- Led by Michal Juhás and Josef Nevoral
- Follow-up meeting four weeks later
From everyday AI chat to advanced skills and agents
4–8 hours · Admin, marketing and sales
- Content built around what slows your people down day to day
- Custom AI skills and assistants for the tasks that keep repeating
- A prompt library tailored to your company, yours to keep
- Materials and examples in English
- Led by Michal Juhás or Andrej Winter
- Follow-up meeting four weeks later
Add-ons, priced by scope
Every company is at a different point, so we price these after the call, once we know what we are talking about.
Tailoring the content
Three to four extra days of preparation: a questionnaire for managers, tasks built on your own processes and data, and prepared repositories. The base price includes two to three hours of preparation.
More challenges and use cases
Additional tasks, if you want to cover more departments or go deeper.
A second instructor on site
For larger teams we split the group and both of us move between them while people build prototypes.
Follow-up meetings
Further online sessions where we work through whatever got stuck in the meantime.
Josef and Michal ran a hackathon for our engineers. Going in, they knew very little about agentic engineering. By the end of the day they had AI autonomously building a new feature, from the ticket all the way to a merged PR.
Martin Valovsky Defence Tech & Brain Health Entrepreneur, SEC Technologies The workshops are led by Michal Juhás. This is how the people he has taught online rate him:
- ⭐ 4.7 out of 5
- on Trustpilot, from 198 reviews
- 1,500+
- reviews on Udemy
- 200+
- webinars delivered
You pay for work delivered
Thirty percent on order, the remaining seventy after the workshop. The price is agreed up front and does not move, however many people turn up. Three days before, we run a technical dry run so we do not spend the day sorting out access, and four weeks later we come back for the follow-up.
The repositories, prompts and materials stay with you, even if you do not continue with us.
What is not included
To avoid misunderstandings: the workshop does not cover deployment to production, connecting to your live systems, hosting, or further development and maintenance of applications. If we agree on that later, we handle it as separate work.
Prices exclude VAT. Travel costs are not included and are billed separately.
Before you decide
The questions companies ask us most often when they are weighing this up.
Why can we not just do this ourselves from YouTube?
Some of it, certainly. There is no shortage of videos. The problem is usually elsewhere: you spend hours working out which parts are even true, and then the team starts twenty things and finishes none. The workshop is not about information, it is about priorities. We pick what makes sense to do first in your company, and build it the same day.
How does this pay for itself?
We are not going to promise you percentages. Every company is at a different point and we are not inventing numbers. Here is a real comparable instead: one estate agency we worked with paid €15,000 a year for a commercial CRM that still did not do what they needed. We built them their own. On the call we go through what the processes AI could take off your plate are costing you, and whether it adds up in your case.
We got burned by consultants once. Why would this go differently?
Because you can check what you are getting. The price is agreed up front and does not move. Thirty percent on order, the rest after the workshop. We also write down what is not included, so there are no surprises at the end. And instead of an eighty-page strategy, you leave with things that work and that stay with you.
How do I justify this to the board or the owner?
We send you material you can forward on: the workshop agenda, the list of processes we would work on, and what your people get out of it. After the workshop you get a summary of what the team built. For defending a budget, that is usually more than enough.
What happens to our data?
We work on invented or pre-approved data. No company or customer records without your consent, and no access keys in the code. During the workshop we also set the rules for what you can and cannot send to public models.
We are not a big company. Is this even for us?
Most of the companies we work with are small and mid-sized, from an estate agency with fifteen agents through restaurants to development teams. AI today is cheaper and more accessible than ERP or CRM ever was. Where it does not add up, we will tell you on the call.
Our team is not technical enough for this.
For the management workshop and the one on skills and agents, no programming is needed at all. For vibe coding and agentic engineering we go through where the team stands beforehand, and if git is new to some of them, we add a short preparation session before the workshop.
The fastest next step
A 15-minute call
Fifteen minutes, so we both know whether it is worth talking further.
- 01 You tell us what slows your team down most.
- 02 We tell you what companies that already solved it did about it.
- 03 We agree which workshop makes sense for you and what we would do first.
You get the most out of it if someone who decides on budget is on the call, or the head of the department this affects. If none of the workshops makes sense for you, we will tell you. You still leave with the processes we would look at first if we were in your position.