I build this for clients, then teach it here
I am Eduard Vlase, a UK software engineer, ten years into the trade. One of the builds I run for a client reads every enquiry that lands in a shared inbox and writes the row into their CRM. I built the AI Automation Hub to teach that same work, which is finding where AI would make a business money, building that part, and getting paid for it.
- Based in
- United Kingdom
- Trade
- Software engineer, ten years shipping software
- Client work now
- AI automations for businesses: lead gen, onboarding, content, support
- Runs
- The AI Automation Hub on Skool, the YouTube channel and a consulting practice
The question I kept getting
I build AI automations for businesses, and very little of it looks impressive from the outside. An onboarding pack that assembles itself without anyone opening a template, a proposal that drafts itself from the notes of the first call, a support inbox that sorts into queues before a person sees it.
Along the way the same question kept coming back: "Can you teach me how to do this myself?"
There are plenty of AI courses out there written by people who came to this a year ago, and I had no interest in adding another one. What I wanted was a place where whoever is teaching you has built, shipped and sold this for a living and is still doing it for clients in the same week the lesson gets recorded. So the Hub has structured courses, templates you can copy into a client project, live calls where we build together as the courses go up, and a forum where members post what they are working on.
It is aimed at engineers, business owners and people changing careers who want to come away with something running.
Four things I will not compromise on
Partner with a business rather than selling it software
Nobody wakes up wanting an automation. They want more customers, more revenue per customer or lower costs. So the first job is always the diagnosis, and a build that does not come out of one is a guess with an invoice attached.
Build the small version first
A client is better off with one workflow running in a fortnight than a platform in six months. The small version tells you whether the problem you diagnosed was the real one, and it is cheap to throw away when it was not.
Foundations before tools
Tools are replaced every few months, but how a model behaves, how data moves and how systems talk to each other changes far more slowly. Learn that layer and most new tools take an afternoon to pick up from their own free academy.
It is done when it survives the edge cases
Anything can be made to work once on a clean input. I test the happy path, then the inputs nobody wants to think about. On a classifier that means running it against a few hundred records I have labelled by hand and reading what it gets wrong. It goes out with monitoring, so the client hears about a failure before their customer does.
Work with me
If you run a business and want this built for you rather than taught to you, my consulting practice and case studies live on my own site.
Learn to build it yourself
The Hub is where I teach the whole path, from learn through build to sell. Join, post your intro and open the first course.
Free while the first courses are being recorded. When it does become paid, one email inside thirty days gets you a refund.