We combine short explanations with hands-on work and review the results together. In the first conversation, we discuss roles, experience, available tools and two or three typical tasks. Then we recommend a course or tailor an in-house workshop.
Build and test AI agents. A day of in-house training for teams who know the basics. Learn how agents and chatbots differ, then build a simple agent and check its responses and tool calls. We agree on tools in advance.
Go deeper with Claude and Microsoft Copilot. A day of in-house training for teams already using these tools. Practise research, presentations and spreadsheet work with approved examples. Build a repeatable workflow with clear review steps. We can discuss hybrid attendance when planning the session.
For a proposal, tell us the team size, experience, available tools and two typical tasks. Request an in-house workshop
Read our guide to Article 4
Build shared knowledge and safer ways to use AI
Some people on your team may already write prompts while others haven’t started. Leaders need to evaluate investments and risks. IT, privacy and security teams need rules that people can understand and use every day. Training helps turn individual experiments into a shared way of working.
A general overview can explain the terms. Hands-on training should also help your team answer these questions.
Which systems are approved for each task?
What data can people enter, and what should they keep out?
How do you spot inaccurate or unsupported output?
When should you check sources, ask a specialist or involve a control function?
What responsibilities belong to managers, business users and technical teams?
How will you support and update new ways of working after training?
We start by discussing who needs training, which systems they use or plan to use, what they already know, and which tasks, rules and records matter. This shapes the course. If usage is unclear, it may make sense to inventory your AI systems or clarify who is responsible before scheduling training.
Who the courses are for
Start with a shared foundation, then build skills around what each role is responsible for.
Audience | Typical questions | Training focus |
|---|
Employees and occasional users | What can generative AI do? What data and tasks are allowed? How do I check the result? | Fundamentals, safe use, prompting, quality checks and escalation |
Marketing, sales, service and administration teams | How can I turn a recurring task into an AI workflow with clear controls? | Role-specific exercises, sources, brand and quality rules, and approvals |
Executives and decision-makers | Which applications are worth pursuing? What decisions, risks, responsibilities and investments follow? | Strategy, prioritising use cases, governance, expected benefits and managing change |
Developers, data and IT teams | How should we design models, APIs, RAG, permissions, evaluation and operations? | Architecture, integration, security, testing, deployment and monitoring |
Privacy, security, legal, compliance and HR teams | What information and checkpoints do operational teams need? | Data flows, policies, records, incident reporting and coordination across teams |
Choose modules that match the work. Technical teams also need to understand what data and processes their organisation has approved. People who review outputs can focus on the checks and rules they use every day.
What participants should be able to do
We agree on learning goals you can assess with realistic exercises. Depending on their role, participants can practise the following.
Explain what their AI tools can and can’t do in their own words.
Use internal rules to distinguish approved, sensitive and prohibited inputs.
Write a structured brief with context, an objective, the required format and review criteria.
Spot hallucinations, missing sources, bias and poor candidates for automation.
Review, revise, discard or escalate AI output when appropriate.
Distinguish the responsibilities of individuals, business teams, IT and control functions.
Identify suitable use cases and recognise when the cost of an error is too high.
Use documented approval and reporting routes in their own work.
Learning depends on prior knowledge, time to practise, group size and the support people receive back at work. We agree up front on the skills your chosen format can realistically cover.
Courses and formats
Choose a shared foundation or a course for a specific role. Each course card and page lists its duration, price, group size and scope. We confirm the details against your needs before making a proposal.
Choose an on-site workshop, an interactive online course or a series of sessions. We can combine an introduction with separate role-based modules and later office hours or follow-up sessions. The format depends on the learning goal, group size, experience, confidentiality and time available.
A short session gives everyone a common starting point. A full workshop leaves time for exercises and questions. A series lets people try what they learn on the job and bring their experiences to the next session.
Training around your tools, tasks and rules
For in-house training, we review what needs to change for your team and how you work.
Systems include approved AI tools, company accounts and the features you use in local or hosted setups.
Tasks come from anonymised examples relevant to the people taking the course.
Data rules cover internal classification, permitted inputs, storage and approval processes.
Quality checks include verifying sources, a second review, brand rules, specialist approval and escalation.
Governance covers owners, contacts, incident reporting and use of unapproved tools.
Language and depth reflect prior knowledge, terminology, technical requirements and accessibility.
We can model sensitive tasks with anonymised or synthetic examples, or use an approved training environment. If a tool or policy is still undecided, we record the open question so the team knows what needs to be resolved.
How we prepare, teach and follow up
We plan what people need before the session and how they’ll apply it afterward. A typical programme has five steps.
Agree on goals and roles. We identify experience levels, systems, relevant use cases and decisions people will need to make.
Review the working context. Your team shares approved policies, information about tools and anonymised examples. We refer open legal or security questions to the people responsible.
Agree on the curriculum. The proposal records the learning goals, modules, exercises, format, group size and equipment.
Run the training. We alternate short explanations with exercises, discussion and examples from your work. We document questions outside the scope for follow-up.
Plan how to apply the training. Depending on the scope, participants receive materials, exercise outputs and next steps. A follow-up can cover what they tried, recurring errors and changes their tools or policies need.
For a larger rollout, start with a pilot group to check examples, language and technical requirements. We use that feedback in later sessions, especially when departments use the same tools in different ways.
What training records can show
We can record the agenda, learning goals, materials and their version, date and attendance. An attendance certificate records who took the agreed course. Short questions, scenarios or feedback can show which concepts came across and where people need more practice.
Attendance shows participation. Ongoing competence, AI system legality and organisational compliance require their own evidence. A knowledge check captures what someone knows at that moment. Useful records also connect training to the audience, systems, goals, content and planned updates.
AI literacy and Article 4
Article 4 of the EU AI Act leaves room for AI literacy measures tailored to people’s roles, experience and systems. Training can support those measures alongside internal rules and regular review of AI outputs. Organisations can choose the right mix of activities and keep their own records.
The European Commission’s AI literacy FAQ explains how to adapt measures to your organisation. Our guide to AI literacy for businesses explores the topic further.
How Carano trained 70 people
The Carano case study covers AI training for 70 people in development, customer success and leadership. The wider programme included guided evaluation of specific tools and a data protection strategy for long-term AI use.
The project shows how training across departments can support tool evaluation and fit with company policies. Assess the content and results for your own teams and work. Productivity, adoption and compliance each need evidence from your organisation.
Where should your team start?
If Article 4 prompted your search, start with the systems, roles and tasks involved. A fundamentals course can give people a shared foundation. You may also need role-specific modules and internal rules. In the first call, we’ll discuss whether an existing course fits, needs adapting or should follow an inventory of your systems.
What needs to sit alongside training
Training builds understanding, gives people time to practise and creates a shared vocabulary. Set up these measures in your organisation as well.
Keep a list of AI systems and assign clear owners.
Put data, identity and access controls in place.
Approve rules for use and make escalation routes easy to follow.
Have specialists review AI outputs where the work calls for it.
Carry out required privacy, security or legal assessments.
Evaluate, monitor and operate each AI system.
Review your approach when systems, purposes or risks change.
If these foundations are still taking shape, start by clarifying AI strategy and governance. Training works best when your tools, rules and processes support what people learn.