I'd rather start by understanding your business than selling you a package. We look at your workflows together — then, only if there's a real opportunity, talk about what's worth building.
20 years building commerce systems at Skype · Microsoft · Bolt
Background in GDPR, ISO 27001, PSD2 and PCI compliance — happy to dig into the data and security questions when they come up
Most AI projects that go wrong, go wrong in a few repeatable ways. Here are the situations where I'd probably tell you so up front — so we both save the time.
If your team has people who can build this, they should — they'll know your context better than I ever could. Me layering on top usually creates friction and an eventual rebuild.
Happy to do a half-day opportunity workshop to point them somewhere useful. But I shouldn't be the one implementing.
If the real outcome is a board slide, a press release, or an investor talking point — I'm not the right person. Those projects don't have a workflow to anchor to, so they drift.
If the goal is a working automation that earns its keep, that's the conversation I want to have.
If invoice approval takes 11 days because three people aren't sure who signs, AI just automates the confusion. Faster, more expensive.
We'd start by mapping the process. Sometimes my honest first take is "don't automate yet — here's what to fix first." I can help with that conversation too — it's just a different one.
Every engagement starts with a free intro call to understand your business. No prepared pitch — just a conversation about your workflows. From there, three possible ways to keep going. You're never locked into the next step.
For 1–10 person teams with one specific workflow on their mind. I look at it end-to-end, write a one-page recommendation, and walk you through it on a call. If it's simple enough to DIY, I'll tell you how.
For 10–50 person teams who want a fuller picture. I map your operation, rank the top 3–5 places where AI honestly pays off, and write it up. If we go on to build, the fee is credited toward the project.
If the audit or pilot points somewhere real, we build it. Fixed price, fixed scope, 2–8 weeks typical. A 30-day support window after launch.
Early days, so I'm taking on a few projects at a time. Honest free first call is the right starting point regardless of which path makes sense afterward — or if none of them do.
These are the patterns I see most often in non-tech SMBs. The numbers below are industry-typical ranges — your real numbers depend on your specific workflow, which is what we'd figure out together first.
I've spent the last 20 years building engineering teams behind systems most people in Europe have used without thinking about it.
I joined Skype in 2005, writing the payment integrations that let people buy SkypeOut minutes with cards, PayPal, and Skrill. Later, as Principal Group Engineering Manager at Microsoft, I was responsible for Skype's Commerce and Monetization work — billing, payments, taxation.
In 2018 I joined Bolt as Director of Engineering, Commerce — the platform behind rides, food, and scooters. Different scale, similar problems: how do you make complex transactional systems boring and reliable.
This past year I kept hearing the same thing from friends running small businesses: "everyone says we should be doing something with AI, but we don't know where to start, and the people pitching us don't seem to understand our actual business."
The unglamorous questions I spent 20 years asking inside big companies — who owns this process? what does the data look like on a bad day? what happens when it breaks at 2am? — turn out to be exactly the questions nobody is asking before AI shows up in small businesses. So I'm starting here.
Talunik is early. I'm working with a handful of friend companies first, learning where I'm genuinely useful and where I'm not. If that sounds like a conversation you'd find useful too, get in touch.
That happens — and when it does, we tell you, before you've spent another euro. You still get the full workflow map and the written report, which have value on their own. If we recommend not to proceed, the audit fee isn't wasted; it bought you the answer.
Data safety isn't something I picked up recently. I've spent years working inside GDPR, ISO 27001, and compliance frameworks at Skype and Bolt — at scale, where getting it wrong had real consequences. Every engagement uses EU-hosted models where possible, keeps sensitive data out of US services unless you explicitly opt in, and documents every data flow. GDPR is non-negotiable.
Process Audits typically start 2-3 weeks out, then take 5 working days once kicked off. Small Business Pilots often start within a week.
Usually no. Most builds run on mid-tier SaaS tools (€20-200/month) plus usage-based AI API costs. Typical monthly cost after implementation: €50-500/month depending on volume.
That's the normal starting point. We explain everything in plain language, show live demos, and make sure your team is confident before we leave. No acronyms required.
Probably yes. Common SMB stacks we connect to: Google Workspace, Microsoft 365, Xero, HubSpot, Pipedrive, Zapier, n8n, Make, Shopify, WooCommerce, Slack. If you use something unusual, the audit tells you if it's workable.
The Small Business Pilot is designed for 1-10 people. Below that (solo operators), we usually recommend DIY tools — and we'll point you to them for free.
Yes, starting 2026 — monthly retainers to maintain AI tools we've built. For now, every implementation includes a 30-day post-launch support window.
Book a free 60-minute intro call. No prepared pitch — just a conversation about your business, the workflows that hurt, and whether AI is honestly part of the answer. If it isn't, I'll tell you.
Monthly at most. Real case teardowns, never a sales pitch. Unsubscribe whenever.