Do I need to know what I want to automate in advance?
No. Describe where work repeats, causes delays or produces errors. We choose the right type of solution afterwards.
Frequently asked questions
40 practical answers about cost, integrations, data, law and long-term support.
Getting started with AI
What happens in the first consultation?
You send your website and one problem in advance. In up to 60 minutes we review the workflow, tools, desired result and major risks. I then recommend the first practical step.
Does an AI audit commit me to further work?
No. The audit delivers a recommendation and a next step. You can implement it with me, internally or with another supplier.
How does the automation assessment work?
You enter your industry, company size, process frequency, time per case and a few operating details. The tool estimates current workload and whether the process is worth investigating. It is not a guaranteed saving or ROI calculation.
Do I need to provide an email to see the result?
No. The result appears immediately. Email is optional and is used only to send the PDF and practical follow-up when you expressly consent.
Does AI make sense for a smaller company?
Often it matters even more. A smaller team handles routine at the expense of sales, customers or the owner's time. Workload, not company size, determines the opportunity, so we start with one process with the strongest benefit.
What can AI do in my industry?
The type of work matters more than the sector. AI helps where people repeatedly read, copy, retrieve or monitor information: enquiries in construction, orders in manufacturing, bookings in services or catalogue support in ecommerce.
Our data is in spreadsheets, email and people's heads. Can we start?
Yes. Most companies start there. We locate the information, consolidate only what the first workflow needs and document the decision knowledge currently held by people.
Should we wait until AI settles down?
No. Start with one small workflow and a measurable outcome. The rules, data and team experience remain useful even when the underlying model changes.
How can AI help me personally as an owner?
An assistant connected to approved company sources can prepare meeting context, retrieve information, draft decisions and turn a transcript into a summary, tasks and a follow-up draft.
Pricing and return
How much does an AI audit cost?
The initial 60-minute audit is free. A broader process audit with a detailed deliverable starts at approximately €2,500 excluding VAT. Scope and price are confirmed in advance.
How much does an AI agent or automation cost?
A focused automation starts at approximately €5,000 and an AI agent at €8,000 excluding VAT. Final pricing depends on steps, data, integrations, testing and operational risk.
How does invoicing work?
Smaller engagements may use one invoice; development can use a deposit or milestones. You receive exact payment terms and price before work begins.
What will the solution cost each month?
Typical tool and model costs range from €15–€60 for a simple automation, €60–€200 for an agent and €200–€600 for larger or voice-based solutions. You pay providers directly. Optional maintenance starts at €120 per month.
Compare maintenance plansHow quickly will it pay back?
Focused automation often targets a 3–6 month payback, but the answer must use your real workload and cost. I calculate it before implementation and separate measured savings from assumptions.
Can grants fund an AI project?
My services are not grant-funded. I design projects to make sense without a grant: one workflow, clear scope, price and measurable result. Use a specialist grant adviser for a larger funded programme.
Why build a custom solution when ChatGPT or Copilot is cheap?
General assistants are excellent for ad-hoc writing and summaries. A custom solution has a trigger, defined permissions, company systems, consistent steps and approval points. The two approaches often complement each other.
Technology and integrations
Can you connect to the tools we already use?
Most projects build on your current tools. I verify the available integration options before proposing a solution.
Which tools do you use?
Depending on the workflow, I use OpenAI, Claude, HighLevel, Supabase, Netlify, Make and standard APIs. Tools are selected by function, data, cost and operating constraints rather than by brand.
Can you work in Czech and English?
Yes. Consultations, documentation and interfaces can be prepared in Czech or English.
Can you connect POHODA, Money S3, Helios, ABRA or Shoptet?
Yes when the relevant edition and licence provides a usable API or reliable export. The exact route is verified before the proposal because product editions, modules and permissions differ.
Does AI handle Czech well?
Yes. Current models understand Czech and Slovak, including conversational language, spelling mistakes and missing diacritics. Tone and terminology are configured for your company.
Can AI answer customers outside business hours?
Yes. A chat, email or voice assistant can answer common questions, record a request or booking and hand a summary to the right person. It identifies itself as AI and offers human hand-off.
24/7 AI assistantHow will this work with our IT administrator?
We collaborate with your IT provider rather than replace them. We review systems and access rules together, document the solution and hand over access. General IT remains with them; our team looks after the AI solution.
Data, safety and law
How do you handle sensitive data?
We define which data the solution needs and where it may be processed. Sensitive information receives explicit access, storage and review rules.
How do you approach GDPR?
We define purpose, scope and data flow, minimise access and storage, and document the services involved. The controller or a legal specialist should confirm legal basis and high-risk processing where required.
Can AI replace employees?
AI can remove parts of routine work, but technology should not decide staffing. I measure the workflow, quality and accountability first. The goal is to release capacity, not promise wholesale replacement.
Can you guarantee visibility in ChatGPT or Claude?
No. I can improve crawl eligibility, indexing, content, trust and measurement. Source selection remains controlled by each service and query.
What does the EU AI Act require when we use ChatGPT or Copilot?
Implement proportionate AI literacy measures, clearly disclose interactive AI, avoid prohibited practices and assess higher-risk uses such as certain recruitment and worker-management systems. GDPR still applies. This is general information, not legal advice.
Our people already use ChatGPT independently. Is that a problem?
The risk is what they enter into personal accounts. Establish the actual use, move suitable work to managed company accounts and create simple rules for personal data, contracts and internal figures.
Who is responsible if AI makes a customer-facing mistake?
The company remains responsible for what it approves, sends or promises. Prices, discounts, complaints and contracts therefore stay behind human approval by default. I correct configuration defects I caused within the agreed scope.
Delivery and support
How long does implementation take?
It depends on scope, integrations and data availability. You receive a concrete timeline before work begins.
What happens after deployment?
We test the solution on real cases, provide documentation and train the people who will use it. Ongoing support is agreed according to need.
Who owns the result?
Ownership of code, accounts, data and documentation is agreed in the proposal or contract. I prefer client-owned accounts and documented hand-over.
What if the solution does not work as specified?
I distinguish a defect, missing input and a scope change. Defects within the agreed scope are corrected; new functions or material changes are priced separately.
Can we work remotely?
Yes. I normally use video calls, shared documentation and controlled access. In-person work is also possible by agreement.
What if you are unavailable?
You do not depend on one person. I work with specialists who can cover maintenance, fixes and development. Solutions run on client accounts where possible and are handed over with documentation.
What if employees resist AI?
Resistance often comes from fear of mistakes or replacement. Practical training makes the boundaries clear, keeps people responsible for consequential decisions and teaches safe data handling.
Practical AI trainingHow much time will implementation take from our people?
Usually a few hours in total: up to 60 minutes for the first consultation, 1–2 hours to describe the process, 1–2 hours for testing and a short training session.
How do I recognise a good AI supplier?
Ask for verifiable work, client-owned accounts, operating costs, test methods, human approvals, cover arrangements and documentation. Be cautious when someone promises full automation without people or clear boundaries.