AI implementation - A recommendation is not yet a change. A working solution is.

Who this is for

• You already have audit findings or a well-described process and want to start with one use case.
• Your team uses AI ad hoc, everyone their own way, and the results are uneven.
• Company knowledge is scattered across drives and inboxes, so the tools answer in generalities.
• You want the solution to stay in the company even when someone leaves.

How it runs

1. 01

   Solution design

   before any building

   I define exactly how it should work: what goes in, what comes out, who uses it and at which point in their work. I choose the tool and justify it, including subscription cost. I also state what the solution will not do - before I build anything, you know what you are getting.

   Result: A designed flow and a chosen tool

2. 02

   Data and knowledge base

   usually the longest stage

   I organise the material the solution will draw on: service descriptions, price lists, procedures, decision history. I unify the format and remove contradictions and outdated content. Without this step AI answers in generalities.

   Result: Organised sources and a list of gaps to fill

3. 03

   Build and configuration

   my own work

   I configure the tool, write and test the prompts, and prepare templates and ready scenarios for typical tasks.

   Result: A working draft ready for your team to review

4. 04

   Testing on real tasks

   on a sample of your work

   We test the solution on your real tasks and compare the outcome with the starting point measured beforehand. I collect feedback from the people who will use it and make corrections.

   Result: A test result compared against the starting point

5. 05

   Instructions and handover

   at the close

   Instructions written for roles, not for the tool. Templates and prompts to copy. Rules for checking answers before use, and a list of situations where the tool should not be relied on.

   Result: Your team works without me

What you get

Common questions