I build AI brains for teams.

A new way of working with LLMs like Claude and ChatGPT. Out of the box, those models know very little about your company, your team or your clients, so so they assume, and most of the time they assume wrong.

An AI brain is built on three things:

• Your company background: what you do, who you sell to, who your clients are, how you price and how you sound.
• A connection to the tools your team already uses: Slack, Teams, Gmail, Calendar, Drive, your CRM, so it can pull the relevant files and data and actually work with them.
• Skills: the tasks your company performs, written down once: generating and sending offers, analysing revenue, automating marketing.

Those three parts, the background knowledge, the tool connections and the skills, are what make an AI brain, and they come together through workflows and agentic workflows. Once your knowledge is structured and shared, everybody can use AI the right way instead of poking at a chatbot with shallow prompts.

No, and be sceptical of anyone who says otherwise. The hype has done real damage here: LLMs and agents get sold as the answer to every problem, which sets teams up to be let down by tools that are genuinely remarkable.

What they actually are is very powerful instruments. Used well, they take the mundane work off your team, produce real assets, and can move revenue. But power is not judgement. They need precise, responsible guidance: the inputs, the rules and the priorities that come out of your vision, mission and values.

An AI brain will not fix poor leadership. It scales the direction you give it, faithfully, including a bad one.

Which is exactly why it is worth building properly. When the repetitive work stops eating the week, people stop spending Tuesday copy-pasting between tabs and start making things that did not exist before. Nobody is bored, the interesting problems finally get attention, and you get breathing room for innovation instead of a backlog of admin.

So set the expectation honestly: this is a daily tool, not a miracle. A super-powered calculator; an intelligent spreadsheet that can act on what it finds. Nobody ever expected the spreadsheet to run the company, but it completely changed what a small team could get through in a week. An AI brain is that shape of change, for everyone.

In the Fractional AI CTO Service, Todor works alongside your CFO, Sales and Product people to capture and structure the organization's scattered knowledge, covering processes, pricing, sales motions and product know-how, and turn it into a company brain the whole team can use. It's delivered with an extended team including data scientists, and the focus is always on efficiency, process optimization and the segments that actually increase revenue, not technology for its own sake.

The AI Literacy seminar is €1,800 flat for 2 hours, half-day session for teams of up to 10 people or a leadership circle. One price whether it runs remotely or on-site; travel and accommodation are added at cost for on-site delivery. All prices are net, excluding VAT.

Every workshop this season also includes an AI hacking quick-sheet, a starter prompt with guidelines for your AI Brain, the Ethical AI usage guidelines, yours to keep and reuse after the session, plus a bonus 30-minute 1 : 1 session two weeks afterwards.

Running a whole cohort rather than one team? The same workshop is priced per cohort, from €1,800 for the session up to €4,200 as Cohort Partner. See the cohort partner options.

It's not a lecture. It's a live, example-driven demonstration run on a real company, with laptops open rather than slides playing. Using LLMs like Claude Code, we show skills being built, quotes automated, the brain repository connected to real tools, designs generated, deals closed through outreach sequences, and revenue analyzed. Real data, real output, including the parts that need a second attempt.

It runs in two modules of two hours. Module 1, AI literacy, is the seminar and live demo: 30 minutes of foundations (agents, connected tools, skills and workflows in plain language), 60 minutes of examples and showcases, and 30 minutes on the EU AI Act and guardrails: security, customer data and GDPR, approval gates, and what not to automate. That module can also be booked on its own as a 2-hour seminar or webinar.

Module 2, hands-on, is optional and can run the same day or a separate one: 30 minutes setting up your tools and mapping what is realistically possible, 60 minutes teaching the model your company knowledge and building the first skill and workflow, then 30 minutes of Q&A and refinements. It needs a Claude Code subscription (or another LLM agent) in place before we start.

Throughout, we discuss how each example maps to your business, and Q&A runs across the session rather than being saved for the end.

Both. Workshops and engagements run remotely, or on-site anywhere in Europe, at the same price. For on-site sessions, travel and accommodation are added at cost on top of the quote. The default working language is English.

Remote works better than most people expect, because the session is screen-driven either way: you are watching a real company's tools being operated, and your team builds on their own machines. On-site tends to win when the point is getting everyone in one room with nothing else in the calendar, which is usually why programmes and leadership circles choose it.

For larger audiences there is also a fully remote webinar format: two hours live plus a two-hour open Q&A window afterwards, at one flat price whatever the headcount. Details here.

Small teams and startups that want to run like a much bigger company; leadership circles under pressure to "do something with AI" but wary of hype-driven investments; and mid-sized companies with real process and data landscapes ready to build their own AI brain, plus accelerator and incubator programmes running a whole cohort at once.

It does not depend on having engineers. The people who get the most out of it are usually the ones drowning in repetitive work in sales, support, marketing and operations, because the whole point of a shared brain is that everybody can use it, not just the technical half of the team.

If AI feels like it's overpromising and underdelivering for your team, this is for you.

Absolutely, yes. Todor has been through several accelerator programmes as a founder, including IKEA Bootcamp, Maze-X and the BMW Foundation Respond accelerator, so this is built from the inside of a programme, not guessed at from outside. That experience decides the format: what actually moves a cohort forward in one session, and what is theatre.

A founder’s time is the scarcest thing in the building, and the hardest part of the job is changing hats: fundraising in the morning, support tickets at lunch, product by evening. The AI brain is what takes the trivial operations off that list. Every startup gets a private session, a timeline built for its own niche, and mentoring across several touchpoints, so that by demo day the teams are spending their hours on mission, vision and traction while the repetitive work runs itself.

Cohort pricing starts at €1,800 flat for the workshop, with a 30-minute 1 : 1 per startup from €2,800. See the cohort partner options. For larger programmes there is a remote webinar format at one flat price, whatever the audience size.

The Fractional AI CTO Service is priced custom per engagement, because scope depends on the size of your organization, the state of your knowledge and the outcomes you're after. Many clients start with the €1,800 AI Literacy seminar to prove the value in two hours, then move into the full service. Get in touch and we'll figure out the right starting point together.

It is part of the work, not a disclaimer at the end. Five principles run through every workshop and engagement: fairness (removing hidden bias from training data so the system treats all groups of people equally), transparency (systems that stay explainable, so people know how a decision was made), privacy (protecting user data and getting proper permission before collecting personal information, GDPR included), accountability (a human takes final responsibility for what the tool does), and safety (designing to do no harm and to prevent malicious or dangerous uses).

The open challenges get named rather than glossed over: algorithmic bias, the environmental cost of training and running these models, and workforce disruption. Every team leaves with the Ethical AI usage guidelines written down, including what to refuse to automate. This is shared for education, not as legal, security or compliance advice; for anything touching customer data or production, bring in qualified legal counsel and a security professional.

Less than you fear if you are small, and more than a checklist if you are large, because what you owe depends on what you use AI for, not how much of it you use.

A small team using AI for offers, marketing, support or research sits in the minimal-risk tier: no conformity assessment, no CE marking. Three duties still bind you at any size: don’t do the banned things (inferring staff emotions in the workplace is the realistic trap), be transparent when someone is dealing with AI or with synthetic content, and support AI literacy in your team. Fines for SMEs are capped at the lower of the fixed amount or the turnover percentage. The one that catches startups: putting AI into CV screening or hiring decisions is high-risk, and so is credit scoring in fintech.

A large company should budget for the programme rather than the paperwork: an inventory of every AI system and a way to classify them, a high-risk compliance programme where it applies, and the deployer duties that come even when the system is someone else’s, including informing workers and their representatives before a high-risk system goes live at work.

Timing as of today: the banned practices and the AI-literacy duty have applied since February 2025, and the transparency duties since 2 August 2026. The high-risk obligations were pushed to December 2027 and August 2028 by the Digital Omnibus, deferred rather than cancelled. Full breakdown in the EU AI Act section. Orientation only, not legal advice.

Everything is proven on a real company first, and shown live. Todor is Product Lead and CTO at goodbag, a small impact startup that uses its AI brain as a daily driver to overperform with minimal operations, so every example is one that already survived contact with real customers, real pricing and real deadlines, not a sandbox demo.

That also means facing the same constraints you do: a tiny team, no budget for headcount, regulated and sustainability-adjacent data, and customers who notice immediately when something reads as machine-made. The workshop covers what that takes in practice, including a fixed brand identity and a designer’s template so output never looks AI-generated, and the ethical guardrails that keep it defensible: fairness, transparency, privacy and GDPR, accountability and safety, plus the open challenges nobody has solved yet.

The guiding principle is simple: a fool with a tool is still a fool. The tool isn't the transformation; the shared brain, the skills and the structure are.

Yes, and be wary of anyone in this field who claims they didn’t. This site, its copy and a good deal of the tooling behind it were built with AI assistance, using the same brain-and-skills approach I would set up for you. Selling a way of working I don’t use myself would be a poor advertisement for it.

Everything here is human-led. AI assisted; it did not decide.

The workshops are led by a person, start to finish. Every example in them is a real problem a founding team actually hit and solved, including the awkward ones: where automation stopped being worth the trouble, where the ethics got uncomfortable, where legal compliance changed the design. No model produced that experience. It came from doing the work and getting some of it wrong first.

And when you get in touch, you reach me. There is no chatbot on this site, no agent triaging your message and no automated qualification step. Your message lands in an inbox and I answer it myself.

Transparency is one of the five principles in every workshop, and the EU AI Act’s transparency duties are part of the guardrails block. It would be hard to teach that with a straight face and then be vague about my own site.

Yes. Alongside the AI work, Todor offers sustainability-first consulting: mapping direct and indirect impact with real data, building a Theory of Change to measure and scale impact, AI strategy and adoption, and shaping customer experience for 5-star products, grounded in impact ventures like goodbag and collaborations with IKEA, Denner, Nestlé and Recup.