headofai.cz — Fractional Head of AI / CAIO for companies in the Czech Republic
For companies that are dealing with AI
Head of AI / Chief AI OfficerAI is already running in your company.Take control of it.
At leadership level I set the priorities, establish the rules and get your first AI workers into measurable operation. Every initiative has to justify a return or a reduction in risk.
90 days · 3 decision gates · no next phase without your go-ahead
Priorities within 30 days
A clear order of where to deploy AI first.
First pilot within 90 days
1–3 priority AI workers depending on complexity — at least one in measured operation.
A clear owner of the AI agenda
One accountable role on the leadership team.
Board reporting and ROI
Impact on revenue, cost or risk.
When a fractional Head of AI makes sense
AI activity is not AI performance.
Tools multiply, experiments pile up — but nobody runs AI as a company capability with an owner, rules and measured impact. That is where my role begins.
AI without an owner
It springs up department by department, but nobody runs it as a company capability.
→ addressed by step 01
Progress without orientation
What AI can do grows faster than companies can make sense of it.
→ addressed by step 02
Risk without boundaries
Data leaks into other people's models, and with agent outputs it is unclear who stands behind them. Regulation does not create this problem — it only makes it visible.
→ addressed by step 03
Activity without a P&L
What gets measured is the number of tools and prompts — not revenue, cost and freed-up capacity.
→ addressed by step 04
Who this is for
Who this is for — and who it is not for.
It makes sense if
+you can name an opportunity or a risk worth single-digit millions of CZK a year
+you have repetitive work in sales, communication or operations
+leadership knows AI will matter but does not want to buy tools blindly
It is not for you if
−you are looking for a cheap chatbot
−you want an inspiring keynote
−you do not have leadership backing or do not want to measure impact
Who is behind this
Why entrust AI to me.
David Veselý
Fractional Head of AI
I help company leadership take control of AI: set the priorities, establish governance, deploy the first AI workers and measure their real impact.
I am not a tool vendor, nor a theorist who knows AI only from slide decks. I build, fund and run AI systems myself. On the Vesely.ai platform I create and deploy AI workers into production; in the venture studio VeselyLab.com I validate automated business models against the real market, and customer acquisition for OmegaTeam.ai is handled by a team of autonomous AI agents under my supervision.
These three environments are my operating laboratory. In them I test the technology, the economics and the operating model before I recommend anything to a client. What actually works is not discovered in a meeting room but in live operation. That is why I do not build your AI agenda on hype, but on approaches I use myself and fund with my own capital.
I represent the interests of your leadership, not of a supplier. Every initiative has to demonstrate a return, a reduction in risk or a clear strategic benefit. When the numbers do not add up, I will recommend postponing or stopping the project. The result is clear priorities, accountability, measurable deliverables and a board report instead of vague promises.
Head of AI is becoming a key role for steering technological change — and someone in your company will carry responsibility for it. At headofai.cz I help companies fill this role in time, before their AI agenda ends up being run in practice by suppliers, individual employees and tools adopted at random.
Operating principles
01Business first. AI second.
02Evidence, not hype.
03Small pilots. Fast feedback.
04I only build what creates measurable value.
05Every AI system has an owner.
06Simple architectures last longer.
Outcomes
What you get in the first 90 days.
A named owner of the AI agenda
One accountable role on the leadership team with a mandate to decide.
CIPHER Snapshot and 1–3 priorities
A readiness map with an estimated annual impact in CZK — instead of dozens of experiments.
Governance and control over AI
A register of AI use cases, rules of use, an accountability matrix and documented training as one of the measures towards Article 4 of the EU AI Act.
AI workers in operation with KPIs
1–3 priority AI workers depending on the technical complexity of the approved use cases — at least one in measured live operation.
A board report with measured impact
A summary of the impact achieved against target and a recommendation for the next period.
Step 1 is free, steps 2–4 form the fixed 90-day sprint, step 5 is an optional continuation. Click a step for detail.
It makes sense if
+you can name an opportunity or a risk worth single-digit millions of CZK a year
+you have repetitive work in sales, communication or operations
+leadership knows AI will matter but does not want to buy tools blindly
It is not for you if
−you are looking for a cheap chatbot
−you want an inspiring keynote
−you do not have leadership backing or do not want to measure impact
30 minutes · no preparation · directly with David Veselý. We name the problem, check the fit — and say plainly whether it makes sense to continue.
What we do
+I assess your AI use cases across the six CIPHER readiness tensions
+I map the data, processes and responsibilities the use cases rest on
+I quantify the economic leverage of each candidate — revenue, cost, risk
C
Cognitive
Does the team have the judgement to tell where AI helps and where it harms?
I
Informational
Is the problem well defined and do you have data for it?
P
Projection
Can you translate the intent into a specification that can be built?
H
Haptic
Where does AI work — and where must the human and the control stay?
E
Exchange
Who decides, who carries accountability, where are the boundaries?
R
Resource
Are the people, the budget and the smallest sufficient solution in place?
Based on the CIPHER framework (Armanios & Tucci, 2025) — open-access paper ↗.
What you get
+CIPHER Snapshot — a readiness map with a pilot / postpone / stop recommendation
+A prioritised portfolio approved by leadership
What we do
+I establish the rules and responsibilities before agents go into operation
+I define where AI decides and where the human stays
Rules that hold up in an audit
+A register of AI use cases — an overview of everywhere AI runs in the company and what it does
+Rules for employee use of AI — what is allowed, what is not, and with which data
+An accountability matrix — for every deployment it is clear who answers for the output
+Documented, contextual employee training — one of the measures towards demonstrably meeting Article 4 of the EU AI Act
Set up with the EU AI Act, GDPR and internal compliance in mind.
What we do
+I design a vendor-neutral architecture on top of your systems
+I deploy 1–3 priority AI workers depending on the technical complexity of the approved use cases — at least one in measured live operation
+I set up impact measurement and human oversight for every agent
The number and depth of deployments is set after the diagnostic, based on the technical complexity of the use cases and the readiness of your systems. The specific scope, responsibilities and acceptance criteria are confirmed in the contract.
Flagship use cases
Website Conversion Agent
Qualifies visitors and guides them to a booking or a purchase.
↗ higher website conversion
24/7 Voice Agent
Takes calls, qualifies and books meetings around the clock.
↗ fewer missed calls
Back-office Agent
Processes emails, invoices and orders without re-typing.
↗ admin hours back
Increasing revenue
Inbound Lead Agent
Handles enquiries from forms, email and DMs within minutes.
↗ response in minutes
Retention Agent
Runs retention, contract renewals and upsell with customers.
↗ lower churn, higher LTV
Reducing operating costs
Customer Care Agent
Handles support and escalates only the exceptions to a human.
Answers according to policy and triggers HR/IT processes.
↗ fewer questions to HR and IT
I get agents running on top of your existing systems and data — CRM, ERP, email, spreadsheets. No infrastructure replacement, no year-long data project.
I direct the architecture and technology choices in your leadership's interest. The implementation is delivered either by us or by a partner you choose. You pay third-party licences and services directly to them, and I accept no commissions for recommendations.
Ongoing management of the AI agenda as an external Head of AI: the portfolio of initiatives, vendors, adoption and board reporting. This includes adoption and training for your team — people know when to trust an agent's output and when to step in.
The target state is not dependence on me — it is an internal owner to whom I hand over a managed system. We tune the portfolio against the data and extend it with new opportunities as they appear.
Return
Work it out yourself.
Conservative numbers. Substitute your own.
Annual operating benefit
24/7 Voice Agent
10 missed calls a month × CZK 30,000 per job × 20% conversion
= CZK 720,000 of revenue a year
The benefit is the gross margin on that — at 35% ≈ CZK 250,000 a year.
One-off capital released
Cashflow Agent
DSO shorter by 7 days at CZK 50m turnover
≈ CZK 959,000 released once from working capital
The annual benefit is the financing saved — at a 6% cost of money ≈ CZK 58,000 a year.
Capacity freed up
Back-office Agent
2 people × 8 hours of admin per week each = 832 hours a year
At a fully loaded labour cost of CZK 480 per hour ≈ CZK 400,000 of capacity a year
The money appears once you reallocate that capacity, remove overtime, or avoid hiring another person because of it.
Three different economic categories, not three comparable returns.
Illustrative numbers — yours are calculated during the diagnostic on your data. If they do not add up, I will recommend not starting the sprint.
Mandate and pricing
A clear mandate, a clear price.
We only go ahead if the expected annual impact clearly exceeds the investment. When the numbers do not add up, I will recommend stopping the project.
Internal Head of AICZK 3–4m a year · 6 months of recruiting · the risk of the wrong hire
An agency for a single projecthundreds of thousands of CZK · no accountability for the AI agenda as a whole
AI Leadership SprintCZK 750,000 · results within one quarter · a stop-point every 30 days
+1–3 priority AI workers depending on the technical complexity of the use cases — at least one in measured live operation
+Handover to an internal owner
You decide phase by phase. Without your go-ahead, no further commitment arises.
Invoiced in 3 payments of CZK 250,000. After every 30 days you decide whether to approve the next phase — each decision commits you to at most a further CZK 250,000. On full completion the total price is CZK 750,000.
Economic threshold for the sprint: we only start it if the diagnostic on your data shows a realistic potential to earn its price back within twelve months.
I direct the architecture and technology choices in your leadership's interest. The implementation is delivered either by us or by a partner you choose. You pay third-party licences and services directly to them, and I accept no commissions for recommendations.
Not another AI workshop
Let's book a decision call.
In 30 minutes we work out whether your company needs its first AI worker, a 90-day sprint — or to sort out accountability first.
After 90 days the AI agenda has a clear owner, approved priorities and rules for data, permissions and human approval. Depending on the technical complexity of the approved use cases, 1–3 priority AI workers are running with defined KPIs, at least one of them in measured live operation. Leadership receives a summary of the impact achieved and a recommendation for the next period.
A specific amount of savings or revenue cannot honestly be guaranteed, because it also depends on the client's data, processes, people and decisions. I guarantee the agreed deliverables, transparent measurement and a clear decision gate. If the diagnostic does not uncover an opportunity whose expected benefit justifies further investment, I will recommend stopping the project. No further phase is delivered or invoiced without the client's decision. Any agreed deliverable that is not met, I complete at no additional fee.
The CEO or a member of the leadership team is needed at kick-off, during prioritisation and for key decisions. Day to day I work with one internal sponsor and the owners of the processes involved. I do not create a pointless working group or an endless series of workshops. The specific time commitment is agreed before each phase.
All three — in that order. The sprint is implementation with a mandate; the retainer is a fractional Head of AI. And when the AI agenda grows into a full-time role, I will help you find an internal Head of AI, onboard them and hand over a working system — the register, the rules, the agents and the metrics. I do not finish by leaving; I finish by handing over.
I build vendor-neutral: frontier models (Anthropic, OpenAI, Google) are chosen by the task, not by partnership. Company data stays in your systems — agents access it through APIs with controlled permissions, and nothing spills into anyone else's training data. Every agent has a defined point where a human decides, and an owner who is accountable for the output. We use least privilege, minimisation of the data passed on, controlled fallbacks and the ability to roll an action back. No system goes into production if it is not clear who is accountable for its decisions and outputs.
The 90-day mandate costs CZK 750,000 excl. VAT and is invoiced in three monthly payments of CZK 250,000 excl. VAT, each before the corresponding 30-day phase begins. For the first payment you receive a complete decision pack — a map of AI activities, priorities with an estimated impact in CZK, and a recommendation to continue or to stop. External licences, cloud and third-party services are not included unless stated otherwise. The client approves each such cost in advance and pays the supplier directly. I accept no commissions for recommending technology. The fractional retainer is invoiced monthly in advance.
Perhaps you don't yet. This role starts to make sense the moment AI affects revenue, costs, company data, customers or how employees make decisions. Without a clear owner the AI agenda does not stop — its direction is effectively set by employees, IT and suppliers according to their own priorities. The job of a Head of AI is not to add another layer of management, but to unify accountability, stop weak experiments and concentrate investment where it has measurable impact.
Yes, to the extent that genuinely applies to you — and without scaremongering.
Most companies are not providers of a high-risk AI system. What most often applies to you is Article 4 and, depending on how you deploy AI, also Article 50.
Article 4 (AI literacy) has applied since 2 February 2025. Regulation (EU) 2026/1744 — the Digital Omnibus on AI, which entered into force on 27 July 2026 — changed its wording: you do not have to guarantee a specific level of literacy for every individual, but you must still take appropriate measures to support its development. It is an obligation of effort, not of guaranteed result.
The transparency obligations under Article 50 generally apply from 2 August 2026, depending on the company's role and how AI is used. For providers of systems generating synthetic content that were on the market before that date, a transitional period until 2 December 2026 applies to Article 50(2).
The main obligations for high-risk systems apply from 2 December 2027 (Annex III) and from 2 August 2028 (Annex I).
I set up governance at your company primarily for the sake of control over data and accountability. Regulatory compliance is a consequence, not the main reason. I am not a lawyer, and the final legal assessment belongs to your legal counsel — I supply the groundwork, not a legal opinion.
Agents take over repetitive work, not people. Deployment includes training for the team and clear rules on who runs and checks the agents. The goal is a team that works with AI confidently — not one that is afraid of it.
