We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
I build agent systems and custom software on your own data and tools, so the repetitive work gets done without new hires. I'm an engineer who studied people, so what I build gets used.
Three things stop automation from paying off.
The busywork grows faster than you can hire.
The prototype impresses, then never touches your data, tools or edge cases.
People route around them, and the manual work comes back.
Systems that take repetitive work off your team, so the same people handle more.
AI agents that work on your own data and tools, running in production.
In A Nutshell Q&A over long videos, with timestamp citations
Web apps built around your process, with live updates where your team needs them.
Tymeslot Open-source scheduling, built in Elixir
I find the manual work worth automating and build it, with n8n, Zapier or custom code.
80% The Happy Beavers less manual work across the agency
Not sure what to build first, or want your team building with agents themselves?
Products I built on my own, and the automation behind The Happy Beavers.
Open-source scheduling that businesses self-host instead of paying SaaS fees.
AI video search
Ask a long YouTube video a question, get the answer with timestamp citations.
Six years as co-founder and CTO, automating how a marketing agency runs.
Also shipped: Transvexis, AI translation in 39+ languages, used by TikTok and Freedom24. Visit Transvexis
One person carries your project from first call to launch, so nothing gets lost between plan and product.
01
A call about your problem and how your team works today, before any spec.
02
A written offer with scope and price before any work starts.
03
Architecture and production code, by the person who scoped it with you.
04
Deployed, monitored and documented. I stay until your team relies on it.
I start with why you need it and how your team will use it, so the system gets used and the work scales.
6 years as co-founder and CTO
Ships production systems
Top 2% nationally, in an MA in Management
BSc Psychology
From the person who built The Happy Beavers with me
Working with Luka at The Happy Beavers was great for the business and for me personally. He sees the whole picture and still gets the details right. He looked at how our team actually worked, found what could be automated and built it, so eleven of us could run a network of more than 150 freelancers. He also brought the team along: he made difficult problems feel manageable, even interesting. If you want your business to grow without growing the team, Luka is the person to ask.
Liliia Yurchuk
Co-founder, The Happy Beavers
Price, ownership and what happens after launch.
The price depends on what the system must do, what it connects to and how much preparation it needs. After a call about your project you get the price in writing before anything is booked, and we look for an offer that works for both sides.
It depends on the scope. Once we have talked through what the system has to do, you get a timeline in the written offer, together with the price.
You do, once the final invoice is paid. The code can live in your own repository from the first commit, and the written offer sets out the details.
The system ships deployed, monitored and documented, and I stay until your team relies on it. After that you choose: your own developers take it over with the documentation, or we agree ongoing maintenance and changes separately.
Another developer can pick it up. I build on a standard, widely used stack (Elixir and Phoenix or Python, with PostgreSQL), document the system as part of the work, and the code can live in your repository.
An agency gives you capacity, but developers rotate and context gets lost between them. With me, the person who scopes your problem is the person who writes the code, from the first call to launch. If you need several developers in parallel, an agency is the better fit.
A ChatGPT wrapper sends what a user types to a model and shows the reply. A custom agent system works inside your stack: it follows your business rules, reads your own data and carries a task through several steps. For example, it can pull CRM data, check stock, draft a tailored proposal and send it for approval. That is work your team no longer does by hand.
Look for someone with formal business training who has also held a strategic engineering role. I studied psychology (BSc) and management (MA, top 2% nationally), and spent six years as co-founder and CTO of the marketing agency The Happy Beavers. The systems I built there cut manual work by 80% and let an 11-person core team run 150+ freelancers. So every project starts with the business problem and the people involved, and the technology follows.
Have a question about your project?
Tell me which work you want off your team's plate. You get a clear view of what it takes and a straight answer on whether I'm the right fit.
A few lines are enough. I reply myself, usually within a day.
Articles on how agents work, for anyone deciding where to use them
Claude now watermarks everything it writes. I read Anthropic's write-up, the EU law behind it and the research to find out who can detect the mark and what removes it.
A Tencent benchmark ranked Elixir first for LLM code generation. As an Elixir developer, I checked the data and found the ranking overstated and the structural advantages real.
Step by step, how an AI agent plans a task, uses tools and checks its own work, and how to tell which of your team's tasks it can take on.
Used by
Transvexis, the machine-translation product I built, served:
TikTok
Machine translation for TikTok
Freedom24
Machine translation for Freedom24's financial content