AI workshops
AI training for teams: AI workshops built to stick
Hands-on sessions for engineering and operations teams, on your own work. Engineers commit agreed rules to your repo, operations teams build one automation, and a few weeks later we check what held.
For CTOs and operations leads who want working habits, not a motivational talk
- Free 30-minute call with Seif
- Price agreed before you book
- You own what the team builds
HorizonLux AI workshops: AI training for teams, on your own work
By the HorizonLux engineering team · reviewed by Seif Sgayer, Founder · updated 30 Sep 2026
HorizonLux AI workshops are hands-on AI training for engineering and operations teams on their own repo and workflows, designed so the team keeps using AI well and safely afterwards. They are for CTOs, heads of engineering and operations leads at companies of 20 to 300 people whose teams already use AI, unevenly and without agreed rules.
The difference is where the training ends up. Engineers commit the rules they tested to the repo, where coding agents read them in every session. Operations teams build one n8n automation for a real workflow, run on test data, with a person who approves its output. Everyone agrees which data may go into which tool, and a few weeks later we check what is still in use.
The workshops are taught by HorizonLux senior engineers who use Claude Code, Cursor, Codex and n8n every day. HorizonLux is a software and AI engineering company based in Tunisia, founded in 2020, working with clients in the UK, UAE and EU. Every workshop starts with a free 30-minute call with Seif Sgayer, Founder, where the track, the work to practise on and the price are agreed.
AI workshops at a glance
- Company
- HorizonLux, a software and AI engineering company, founded in 2020
- Based in
- Tunisia, with clients in the UK, UAE and EU
- Taught by
- HorizonLux senior engineers who use Claude Code, Cursor, Codex and n8n daily
- Format
- A hands-on session of CONFIRM: typical length per track, CONFIRM: online, on-site or both, in English, then a follow-up check a few weeks later
- Group size
- Up to CONFIRM: maximum people per group, so everyone practises
- You own
- Every rule, prompt and automation the team makes, in your repo and accounts. NDA on request
- Starts with
- A free 30-minute call with Seif Sgayer, Founder, with no commitment
Who teaches our AI workshops: engineers who build AI for clients
We have not published a workshop case study yet, so judge the people who teach by what they build for clients. The free call lets you judge the plan before you pay for anything.
RepairCheck
Client build · GermanyVehicle damage inspection. We built AI damage analysis with OpenAI vision, a copilot that writes to the case record, n8n flows, and an admin dashboard with prompt configuration and audit logs. Auto-estimates stop once an expert claims a case, so a person takes over. The workshops teach the same pattern: an AI step, a check, and a person who decides.
React Native · Expo · Next.js · Postgres · OpenAI · n8n
Read the RepairCheck case studyYour workshop lead
Senior engineerA HorizonLux senior engineer who works in Claude Code, Cursor, Codex and n8n on client projects leads the session and the follow-up check. CONFIRM: name, and two lines on what they build for clients. Seif Sgayer, Founder, runs the free call and agrees the plan with you.
Claude Code · Cursor · Codex · n8n
…how much ownership he takes: he thinks a feature through, raises edge cases before they turn into bugs, and delivers something that actually works… Exceptional clear and consistent communication.
Corporate AI training lags behind everyday use
People adopted AI faster than companies trained them or agreed rules for it, and the output still needs checking. Three published surveys, none of them HorizonLux data.
- 36%
of employees surveyed feel they received adequate upskilling.
Source: Boston Consulting Group, AI at Work, 2026
- 88%
of employees with enterprise AI tools also use personal AI tools for work.
Source: Gartner, 2026
- 66%
of developers cite AI answers that are “almost right, but not quite”.
Source: Stack Overflow Developer Survey, 2025
Why AI training for employees fades: the rules stay in the slides
BCG's rule of thumb (2024): about 70% of AI challenges come from people and processes, not the technology.
When the rules stay in a deck, it looks like this. One developer pastes a log with customer data into a personal chatbot account. Another merges a large agent diff after a quick skim. A third stops using the tool after it was wrong twice. Nobody opens the deck at their desk.
Our workshops put the rules where the work happens. We bring the rules we work by on client code, your team tests them on its own tickets and workflows, and the version that survives is committed to your repo or built into the automation. A rules file is context the agent reads, not enforcement, so anything that must hold every time goes into hooks, permission settings or CI checks.
The five rules shown here are a sample, not a client's file.
Secrets stay out of prompts
No .env files, keys or passwords in any AI tool. Customer data goes only into the tools and accounts the team approved for it.
Plan before any edit
The agent writes a plan and waits. A person reads it and says OK before a file changes.
Tests before the change
The agent writes or updates tests before it touches the code, so the check exists before the change does.
A person reviews every AI-written PR
Small diffs, one concern each. The PR says what the AI wrote, and nothing merges on the agent's word.
A person approves what an automation sends
Each n8n flow with an AI step checks its input and waits for its owner's OK before anything goes out.
What each AI workshop leaves in your tools
Pick one track or combine them. Each ends with files and flows in your own repo and accounts, not a slide deck.
AI coding workshop for engineers
Claude Code training for developers, with Cursor or Codex where your team uses them. On your own repository, the team takes a real ticket from the agent's plan to a reviewed pull request.
- A rules file the agent reads every session
- Hooks or CI checks for rules that must hold
- Permission settings: what the agent may run
- Review rules for AI-written pull requests
Claude CodeCursorCodex
Build workshop for operations teams
Support, finance or operations teams build one n8n automation for a repetitive task from their own week, with an AI step, a check for bad input and a person who approves before anything goes out.
- One automation, built and run on test data
- A one-page runbook for its owner
- The prompts, saved where the team works
n8nClaudeOpenAI
Safe everyday AI use for everyone
How to give an AI assistant the right context, how to check its answer, and which data may go into which tool. Written down with the people who have to follow it.
- Data rules per tool: allowed, test only, never
- A draft AI usage policy your lead finalises
- A dated agenda and attendance record
ChatGPTClaude
How an AI workshop runs: before, during and after
Three of the four steps happen outside the session. That is the part a one-off talk skips.
- 01Free · 30 min
Free call
You tell Seif Sgayer which AI tools your team uses, where it goes wrong and who should attend. The track, the length and the price are agreed before anything is booked.
You get · A suggested track and an agreed price
- 02Before
Prep on your work
We look at your repo setup and workflows and pick the real tickets and tasks to practise on. Exercises use test or non-sensitive data, never production secrets.
You get · Exercises built on your team's own work
- 03Hands-on
The workshop
Your team works in its own tools, with our engineers guiding. After each exercise the team agrees the rule it just tested and writes it where it will be used.
You get · Rules committed, one automation built, data rules agreed
- 04After
Follow-up check
A few weeks later we review real pull requests with your lead, and the automation's runs if it went live: which rules are used, which slipped, what to change. We measure use, not hours saved.
You get · Updated rules and a short note on what held
Sessions run in English, in your team's working hours on request.
Four ways to train your team on AI, and where each one stops
Each option has a place. Here is what each one gives you, side by side.
| What matters | Learning on the job | Free vendor courses | A training company | HorizonLux AI workshops |
|---|---|---|---|---|
| What people practise on | Whatever comes up | Generic examples | Prepared exercises, some tailored | Your repo, tickets and workflows |
| Who teaches | Colleagues, when they have time | Self-paced lessons from the tool's maker | Professional trainers | Engineers who use these tools daily |
| Rules for data and code | Unwritten, different per person | General guidance | Varies by provider | Agreed by your team, written into your tools |
| What stays afterwards | Individual habits | Personal notes | Course materials, often a prompt library | Repo rules, a working automation, a draft policy |
| Follow-up | Informal, if any | None | Often little | A check on real use a few weeks later |
| Cost | Staff time | Free, plus staff time | Per person or per session | Agreed on a free call before booking |
| Suits | Early experiments | Individual basics, as pre-work | Standard sessions on common tools | Teams that need shared rules on real work |
Categories, not named companies. Anthropic and OpenAI both offer free self-paced AI courses. We can suggest one as pre-work, so live time goes to practice.
Questions about AI workshops
Direct answers on price, format, skills, tools, ownership and the EU AI Act.
How do you train employees to use AI well and safely?
Train them on their own work, and write what they agree into the tools they use. That is how HorizonLux runs its AI workshops. People practise on a ticket from their backlog or a real workflow, so nothing needs translating back at their desk. The team agrees which data may go into which tool and who checks AI output. Engineers commit those rules to the repo and CI; operations teams build them into the automation. A few weeks later, the team lead checks which rules are still followed.
What does an AI workshop cost?
The price is agreed on the free 30-minute call, before anything is booked. It depends on the track, the number of people, the length and how much preparation your repo or workflows need. We don't publish a list price, because a session on your own work has to be scoped before it can be priced. You know the full price before you commit, and the call carries no commitment.
How long is a workshop, and how many people can join?
A typical workshop is one hands-on session of CONFIRM: typical length per track, for up to CONFIRM: maximum people, CONFIRM: online, on-site or both. Groups stay small so everyone practises rather than watches. Preparation on your tickets and workflows comes before the session, and the follow-up check a few weeks after it. The date is set on the free call.
Do people need technical skills to take part?
Only for the engineering track. The everyday-use and build tracks are for anyone who uses AI at work, in support, finance, operations or sales, and need no coding: our engineers guide each step of the n8n build. The engineering track is an AI coding workshop, Claude Code training for developers with Cursor or Codex where your team uses them, so participants should already work in your repository and review pull requests.
Which AI tools do you teach, and is our data safe in the exercises?
We teach the tools our engineers work with, and exercises use test or non-sensitive data, never production secrets or customer records. For engineers: Claude Code, Cursor or Codex, which our team uses daily. For automation: n8n with Claude or OpenAI models. For everyday use: assistants such as ChatGPT and Claude. If your team pays for another tool, such as Microsoft Copilot or Gemini, raise it on the free call and we will say plainly whether we can help.
What will my team have afterwards, and who owns it?
You own everything the team makes, and it lives in your repo and accounts. For engineers, that is the rules file, the hooks or CI checks and the review rules. For operations teams, it is the n8n automation, its runbook and its prompts. Everyone keeps the data rules per tool, a draft AI usage policy and the attendance record. An NDA is available on request.
Does this count as EU AI Act training for employees?
It can be one of the measures Article 4 of the EU AI Act asks for, but no workshop alone makes you compliant. Article 4 has applied since 2 February 2025. As amended in 2026, it asks providers and deployers to take measures to support AI literacy among staff and others operating AI for them. It sets no fixed level and has no official certificate. Since 2 August 2026, national authorities can enforce it. You keep a dated agenda, attendance record and agreed rules. We are engineers, not lawyers.
How does it start, and what if we need more than a workshop?
It starts with a free 30-minute call with Seif Sgayer, Founder, with no commitment. You describe your team's AI tools and where things go wrong, and Seif proposes a track and the work to practise on. If you go ahead, we prepare on your repo or workflows before the session. If a workshop surfaces a bigger build, it can become a paid pilot with one agreed deliverable, which you can stop if it does not land.
See what your team would still be using next month
Your team is forming AI habits either way. Spend 30 minutes with Seif Sgayer, Founder, deciding which ones to keep.
Book a Free 30-Minute Call
Tell Seif which AI tools your team uses, how AI-written code and output are checked today, and where it has gone wrong. Seif suggests a track, the real work to practise on and what should end up in your tools. No slide deck and no commitment.
Don’t want a call? Email [email protected]
Book a free call- Your team's AI use reviewed, tool by tool
- A suggested track and the work to practise on
- Length and price agreed before you book
We worked with Seif on the React Native app for our vehicle-appraisal platform. What stood out most was how much ownership he takes: he thinks a feature through, raises edge cases before they turn into bugs, and delivers something that actually works. And then there is one more thing why I wanted to work with Seif after our very first call: Exceptional clear and consistent communication. It is a pleasure to work with him and his team!
Michael EmaschowFounder, RepairCheck
Seif SgayerFounder ·View LinkedInFree AI Workshop Call
30 minutes · Google Meet · Free
No packages, no sales pitch. You leave with a clear first step.
- 30 min
- Google Meet
- Calendly
- No commitment
- The plan is yours to keep
- Built for teams at Tabeebi, RepairCheck and MasterPilot
HorizonLux is an independent company. Claude, Claude Code, Cursor, Codex, ChatGPT, OpenAI, Anthropic, n8n, Microsoft Copilot and Gemini are trademarks of their owners, who do not sponsor or endorse these workshops.