ATC is a specialised AI Act consultancy based in Bulgaria. We help companies operating in Bulgaria and across the EU understand the EU AI Act and comply with it — with concrete services, published prices and support in English, Bulgarian and Russian.
Whether you need an AI Act consultant for a single risk classification or an AI consulting company to support you long term, ATC is built for companies with operations, teams or customers in Bulgaria.
Bulgaria is an EU member state, so the AI Act applies here directly. We follow how it is being implemented locally — which national authorities are designated, what is still missing, and what that means for your obligations.
Many international companies run development centres, support teams or subsidiaries in Bulgaria. We work in English, Bulgarian and Russian and prepare documentation your headquarters and your local team can both use.
We are not a general IT consultancy with AI Act as one line item. ATC is dedicated entirely to AI regulation, working to a published methodology — the ATC Trust Standard, built on ISO/IEC 42001.
Every service has a published starting price. Before work begins you receive a written offer with exact scope, timeline and cost — invoiced by a registered Bulgarian company you can verify in the Commercial Register.
Yes. The AI Act applies across the EU, and we work with international companies that operate in Bulgaria or serve EU customers. Engagements are conducted in English and run remotely.
With a risk classification of each AI system you build or use — unacceptable, high, limited or minimal risk — followed by a written plan of the specific obligations that apply to you and the order in which to address them.
No. ATC combines regulatory and technical expertise. Where formal legal representation is required, we work with partner lawyers.
The EU AI Act is already in force in stages and affects far more companies than expected — not only those building AI, but anyone using it day to day. If a staff member writes emails with ChatGPT, if your site has a chatbot, if you screen candidates with automated software — the regulation already applies to you, even if you haven't realised it.
ATC is the first specialised Bulgarian resource dedicated entirely to this regulation — not a generalist IT consultancy adding AI Act as another line item between cybersecurity and cloud solutions, but a team built specifically and solely around it. When your full attention is on one thing, you understand it more deeply than anyone treating it as a side note.
Behind ATC stands a team combining understanding of the regulatory text with practical technological knowledge — not lawyers reading a law without grasping how an AI system actually works, and not technologists dismissing the legal weight of non-compliance. When legal escalation is needed, we work with a prepared team of partner lawyers ready to take on cases requiring formal representation.
We don't sell fear of a fine — we sell clarity. Every client gets a concrete, written answer to "does this affect me, and if so, exactly what do I need to do" — instead of abstract talk about "compliance" and "risk." We work to our own, publicly documented methodology (ATC Trust Standard), built on the internationally recognised ISO/IEC 42001 standard — not improvised, but a verifiable, written-down system.
The most expensive mistake isn't paying for a consultation — it's discovering too late that you've missed an obligation a fine is already accruing on.
Many business owners we talk to go through the same moment — reading for the first time exactly what the regulation requires, and realising: "This really needs checking. I may have already missed something." That moment of realisation is exactly why ATC exists — so you don't reach it alone, in a panic, once it's already too late for a calm decision.
In practice, employees in many companies already use AI tools at work, often without explicit approval from management. This isn't just a curiosity — it means a company may already be a "deployer" under the AI Act, even if the owner never consciously approved the use of a specific tool.
Whether AI is used openly or covertly, out of a genuine belief in its efficiency or simply because it's faster — responsibility before the regulator remains with the company, not the employee. Management's ignorance is not a legal defence.
Each service has its own page with a full description, process and answers to common questions. You only order what you actually need.
We determine whether and how the EU AI Act applies to your product or business. The service starts with risk classification of each of your AI systems across the four levels set out in the regulation — unacceptable risk, high risk, limited risk, and minimal risk — and continues with a concrete plan of what steps you need to take based on the result.
Learn more →GEO (Generative Engine Optimization) is the new discipline that follows the logic of classic SEO, but is aimed at AI systems rather than a classic Google results page. More and more people no longer search for products on Google — they ask ChatGPT, Gemini, or Perplexity directly, and if your business isn't structured so these AI systems can understand it, it's effectively invisible on this growing channel.
Learn more →Autonomous AI agents (agentic AI) no longer just answer questions — they shop, compare prices, book, and even complete purchases on behalf of the user. Siri, ChatGPT and other assistants already include such features, and analysts expect agents to manage hundreds of billions of euros in online purchases in coming years. If your site or product isn't technically readable and accessible to these agents, you're losing transactions you don't even realise you're missing.
Learn more →Companies with high-risk AI systems must maintain full technical documentation under Annex IV of the regulation — system architecture, a risk management file, system operation logs, evidence of human oversight. The challenge isn't producing this documentation once — it's keeping it up to date as the system changes.
Learn more →Companies that buy and use ready-made AI tools (rather than building them) also have obligations under the AI Act — in the role of deployer. You must vet the vendor, document how you actually use the tool, and monitor whether the purchased system doesn't unexpectedly fall into a high-risk category based on how you use it.
Learn more →The AI Act and the related European Code of Practice introduce new transparency requirements — users must be able to tell when they're interacting with a chatbot rather than a human, and when content (text, image, audio, video) has been generated or significantly altered by AI. The requirement covers marketing materials, customer chatbots, generated images, and many other cases most companies haven't yet considered.
Learn more →The regulation explicitly introduces an "AI literacy" requirement — staff working with AI systems must understand enough about how they function and what their risks are. No technical expertise is needed, but a basic, documented understanding is, one you can show during an inspection.
Learn more →A full-time AI director costs over €200,000 a year — out of reach for most small and medium businesses. A Fractional AI Director is the same strategic role at a fraction of the cost: ongoing, monthly presence at the decision-making level, without the full-time price tag.
Learn more →Three questions. No email, no signup — just an honest answer.
Systems that influence decisions about people most often fall into the high-risk category under the AI Act — with specific, substantial documentation and oversight obligations. We recommend a consultation for exact classification as soon as possible.
See AI Act ConsultingYou likely fall into a category with lighter obligations — most often transparency requirements, not full documentation burden. Still worth a quick check to be sure.
Browse servicesIf you don't use AI yet, the regulation doesn't affect you directly today. But it's useful to understand the basics now — see our glossary of terms.
See the glossaryATC Trust Standard is our public methodology for assessing and building AI compliance — built on top of the internationally recognised ISO/IEC 42001 framework and EU AI Act requirements, with our own additional controls for areas these documents don't yet fully cover: agentic AI behaviour, GEO/AI-discoverability readiness, and practical, accessible application specific to the Bulgarian market.
We do not claim formal equivalence to ISO certification — this is our own, transparently published framework, which you can verify point by point rather than simply take on trust.
The EU AI Act applies directly across all member states, but each country must designate its own national authorities responsible for enforcing it. Here's what has actually been done in Bulgaria, and what's still missing.
The official terms from the regulation, kept exactly as you'll encounter them everywhere — but explained in plain language, in enough depth to actually understand them, not just recognise them.
