AI transformation and executive workshops

Building AIThat SurvivesThe Second Year.Not The Demo.

Year two

Sauhard Dubey. I run AI transformation end to end inside small and mid-sized companies, and I teach senior teams what to ask before the money moves. Munich is home and most of the work sits across EMEA.

Start a conversationThe ways in
Where I learned it

German industry, four companies, four different ways of being stuck.

01Retail

The Price MovedFaster ThanThe Model

Retail taught me the clock. A forecast that's right on Thursday and late on Friday is worth nothing, so the question stopped being how good the model is. It became how fast the business can move on what the model says. Almost everything I design still starts there.

Ways in

Different Problems.The Same Few Decisions.

Some of this is a programme with your team inside it from the first day. Some of it is one room and a lot of direct questions. What they share is where the value sits: in the handful of decisions that look obvious afterwards and cost you a year when they go the wrong way. I take on a small number of each, alongside my work at appliedAI Initiative.

  • End to end

    AI Transformation, Complete

    I take a company from its first open question through to systems people use on an ordinary Tuesday, then hand it to the team who runs it afterwards. Much of that is one decision made well: what belongs in a low-code tool like n8n, and what needs real code underneath it. I build in n8n most weeks, which is exactly why I'll say when it's the wrong home for something. Get that call wrong in either direction and you pay for it every year after.

    For small and mid-sized companies that want it running
  • Strategy

    Where AI Pays, In What Order

    Before anything gets built, someone has to decide which parts of the business AI should touch. The harder half of that is what to buy off a shelf instead, and what to leave alone for another year. I work it out against what you already run and what your people can absorb, and then we cost it honestly, because the pilot is never the expensive part. Inference at volume is, and so is the retrieval layer, and so is whoever is still maintaining all of it in year two.

    For boards deciding where to commit next year
  • Executive education

    Workshops For Senior Teams

    A working session for the people who approve AI spending and later have to defend it. We cover what the models can and can't do, then the two questions nobody prepares for: how you measure impact once something is live, and what to do when a good system goes unused. We spend real time on why something that dazzles in a demo can come apart in its first week against your own data. Your use cases, not case studies about somebody else.

    For management teams that keep the decision
  • Building together

    Technical Cofounder

    Some of the best ideas I hear come from people who know their market cold and have no way to build the thing they can already see. That's a good place to be standing, and it's the one I like walking into. I'd take the technical side, from the architecture through to the early calls that get expensive to reverse. Bring me the idea. We can work out the rest from there.

    For founders who know exactly what to build

One Email.One Honest Answer.

Tell me what you're trying to do and I'll tell you whether AI is the right instrument for it. If it isn't, I'll say so, and you'll have saved a budget cycle. If it is, we can work out which way in fits.

Email meLinkedIn
Or meet the person first, and everything that got built along the way