1. Introduction
We start with a half-hour conversation. Where does your team's time go? Which work is repetition, and which really needs a professional? From that we choose together the first piece of work for an agent.
2. A plan per phase
Every piece of work is a phase with its own plan: what the agent will do, what is done when it is done, and how many hours it takes. You read the plan and approve it before we start. Large projects are split into phases that can be finished in a few weeks.
3. Build and show
We build the agent and first let it work in a test environment, with examples from your own practice. You see what it proposes and why. Your feedback is the next assignment.
4. Live and maintained
Only when you are satisfied does the agent go live. We look after it and adjust it when your work changes. Every month you see the hours and AI costs, and every step the agent takes can be traced.
What we care about
Agents work with sensitive data and with work that carries your name. That is why we keep to four fixed agreements.
- Client data is stored in your own environment, and which AI service sees what is agreed in advance
- Anything that goes to a client or the tax authorities is approved by a person first
- We are not tied to one AI model and choose what fits each task
- Hours and AI costs are visible every month