I built an AI that actually executes for tradespeople instead of just chatting

Hey IH 👋

I'm Timo, solo founder of SalesHookAI (Zera). Building this for German/Europe trade businesses ( plumbers, contractors etc.) — the kind still running quotes on paper, scheduling on a whiteboard, invoices typed by hand.

Most Software on the market have an LLM wrapper only integrated: stateless chat interface bolted onto software, no real access to the system. I wanted Zera to be different, so over the last year I built:

- Agentic RAG — she pulls relevant business rules, customer preferences, and special conditions automatically at every chat start, no need to re-explain

- Persistent vector memory per customer — payment terms, delivery times, rates, VAT categories, stored permanently and recalled automatically

- Function calling across the full system — CRM, quotes, projects, Planbook, time tracking, fleet. "Schedule an appointment with Mr. Schmidt at 2pm" isn't a generated text reply — it actually changes state in the system

- Proactive monitoring — she flags a staffing gap for tomorrow or an unanswered quote sitting for 3 days, without being asked

It's built on Google Vertex AI

Where I'm at honestly: 2 paying customers so far. Bootstrapped, no team, every hour counts — which is exactly why I'm posting here instead of just quietly shipping.

Would love your take:

- If you ran a service business, what would make you trust an AI enough to let it take actions on your quotes/schedule, not just suggest them?

- Does this pitch land, or does the technical depth (RAG, vector memory) bury the actual value for a non-technical tradesperson?

I'd be happy to go into more detail about the architecture or the challenges involved in going it alone to win new clients.