Practical AI training
One day. Documents, spreadsheets, research and first small automation, all hands on.
For teams that want a shared foundation, to try AI on safe tasks and to leave with one use case worth developing.
From first training
to one use case.
A practical day gives the team a shared foundation. From there we follow with verification and adoption of one concrete use case.
BOOK THE TRAINING
AI adoption in the company
Follow-on work on one process, with your data and your rules.
For companies that already know where they want AI. First we measure benefit and risk in limited operation, only then do we propose the rollout scope.
AI on real tasks.
From a pile of files to a first draft
AI pulls the key points from your files and prepares a first version. Sources and conclusions are checked by a human who knows how.
A question instead of a formula
The team says in plain language what it needs to find out. AI prepares the calculation or chart and people check how it reached the number.
A task that stops repeating
Repeated work becomes a procedure with a clear input, checkpoint and output. On training with trial data, in your company with yours.
EUROPEAN UNION · REGULATION 2024/1689Prohibited AI use:
Prohibited AI use:
€35 million,
or 7 % of turnover.
It applies across the European Union and it is the higher of the two amounts. Most other breaches are capped at €15 million or 3 % of total worldwide annual turnover. Since 2 August 2026, AI-generated content must also be labelled and people must be told when they are talking to a machine. More obligations are being added.
A hands-on day.
From brief to check.
Preparation by role
Beforehand we find out what people work with and prepare trial tasks from their field. We do not need your data for the training day.
Everyone works alone
Documents, spreadsheets, research, small automation. The instructor teaches how to sharpen the brief and verify the output.
One priority
At the end we choose with management one use case worth developing. Deployment is agreed separately, as a standalone step.
Data, checks, accountability.
In every procedure.
For every scenario we cover three things: what data may enter a tool, how to check the output and where a human decides. We offer standalone verification of a system where it truly matters.
Data and tools
What may enter which tool, what stays inside the company and when to use a separate workspace. The team knows before they paste anything.
Output checks
Facts, calculations and sources are verified against the original. People recognise tasks where a human always has the last word.
Access and automation
Every repeatable procedure has defined permissions, a checkpoint and a stop button. Unsupervised automation is not built here.
We publish measurements of the Czech web and AI work including methodology.
Our own research.
Open results.
CIAD is a registered institute. Our purpose is on the public register: to protect and educate consumers of AI services, to alert the public to unethical use of AI, and to teach companies how to work with it properly.
That is why we sell nothing off the shelf. We teach how to work with context, files and output checks in a way that survives a change of model, and we verify every procedure ourselves first. It separates what holds up in daily work from an impressive demo.
Security and regulation are part of responsible AI use, but the goal is useful work. The institute is led by its director and statutory body, Zbyšek Chudoba.
8published studies with open data
6free Czech-language briefings
0ties to vendors
News
CZECH ONLY
Articles beyond the English selection and the NIS2 self-test are published in Czech only; see the blog.
Mend.io framework secures AI agents and LLM apps in production
Mend.io publishes a practical framework mapping the attack surface of agentic systems with seven reusable artifacts and runtime guardrails for production deployment.
READ ARTICLEShadow AI exposes companies to AI Act and GDPR penalties
Unauthorised use of public AI tools leaks sensitive data and triggers liability under new EU rules from August 2024.
READ ARTICLEFirst self-replicating AI virus prototype demonstrated by researchers
Toronto and Cambridge researchers show an open-weight model on a single GPU can autonomously find, exploit and replicate across systems.
READ ARTICLEKnowledge centre.
Free, no sign-up.
FAQWhat companies ask
What companies ask
before they write to us.
What is CIAD and what does it do for companies?
+
CIAD is an independent registered institute. We teach Czech companies to work with artificial intelligence on their real work with documents, spreadsheets and repeated tasks, help them adopt a first concrete use case and independently verify AI systems before they receive sensitive data. Alongside that we measure how the Czech web and companies work with AI, and we publish the results. We are not tied to any AI vendor and we sell nothing off the shelf.
What does it mean that CIAD is on the consumer side?
+
The purpose of the institute, as entered in the register, includes protecting and educating consumers of services that use AI and alerting the public to entities that use AI unethically. In practice this means we measure and publish how AI is actually used in Czechia, and we advise companies so that their use holds up before customers too. No vendor pays us, so we have no reason to embellish anything.
What is the EU AI Act and who does the AI literacy obligation apply to?
+
Article 4 of the AI Act, as amended by Regulation (EU) 2026/1744, applies to providers and deployers of AI systems. It requires them to take measures supporting AI literacy according to role, knowledge, experience and how AI is actually used. It is an obligation of effort, not of result: a company does not have to guarantee the level of any individual. The regulation does not prescribe one course or certification.
How does company AI training work and what does the organisation document?
+
First we identify roles, real tasks and the target outcome. Then people spend one day working with AI on tasks from their field, in person or remotely, and at the end we choose with management a first use case to develop. The syllabus, attendance records and role-based rules help you document the measures taken under Article 4. They are not a legal opinion or an automatic confirmation of compliance.
How does independent verification of an AI system work?
+
Together we define the system, data, permissions and abuse scenarios that make sense for your deployment. We test them in a targeted way and hand over findings with documented impact, severity and a remediation plan. After the fix we verify that it holds. The detailed procedure is described on the audit process page.
How is CIAD different from security and consulting firms?
+
We take no money from vendors whose tools we evaluate, and we sell no technology we would then recommend to ourselves. We are a registered institute whose purpose is to protect those who use AI, not a consulting firm with a catalogue. We back our conclusions with our own published studies with data and methodology, so anyone can verify them.
What does the first step of working together look like?
+
In a short message you describe the roles, tools used and the work you want to change. We come back with a proposed format: training, adoption of a concrete use case, or verification of a system. Pricing follows the concrete scope and we quote it on request.
Book training for your team.
Tell us who you want to train and what your people work with. We will reply with a proposed format. You speak directly with the institute.
- office@ciad.cz
- Phone
- +420 607 445 004
- Office
- Příčná 1892/4
110 00 Prague 1