Eight projects with the result on the page, and two that are in delivery now. Each case shows how the work ran before, what we built and what the team sees now.
Every KVVK estimate used the same expert logic, but it still took several hours to turn project files into a reviewable draft.
Now Baltimir starts from a prepared AI workspace and reviews a marked estimate draft about 30 minutes later.
Estimates were in Excel and costs in Merit, so seeing where a project stood meant building a report by hand.
The estimate is loaded once. Every morning the dashboard checks Merit costs against it, project by project.
Lead source, follow-up and deal outcome are now connected, so weak channels can be adjusted faster.
Every call is now transcribed, scored against the playbook, and flagged for coaching while the context is fresh.
Short doctor notes now become review-ready article drafts in three languages, with tone and safety checks built in.
Teachers practised with curriculum-aligned AI assistants for lessons and classroom use; school directors had their own session.
Every deal now has an owner, stage and forecast the CEO can inspect before a management meeting.
A paid first step, calls to past clients and monthly memberships from August, built in weekly sessions over the summer.
Neither has a result to show yet, so there are no result figures here. This is what is already built and what happens next.
Monthly document reminders have run since 1 October 2026. The first real invoices were created in October, and the team checks them before they go out.
Next: October's invoices will show how much extra work is billed on top of the package. Then this becomes a full case study.
The order desk has been in use since August.
Next: Before the quoting assistant is built, we measure how long a quote takes today.
We look at one job that takes too much time right now, and I tell you whether it is worth fixing.
Free, on Google Meet
