AI Adoption & Automation
We build AI tools for your work and automate the tasks your team repeats every day. The goal is not to have adopted AI — it is for the hours to actually go down.
What we take on
- Custom AI tool development — Tools built around how your business actually works
- Automating repetitive work — Look it up, draft it, send it — the loop you run every day
- Connecting existing services — API integrations with Gmail, X (formerly Twitter) and others
- Technical consulting on AI — Working out what can be handed to AI, and what should not be
- Support after rollout — Tuning accuracy, and making sure people actually use it
Contract work: Ichigo AI
An AI work tool we built for a client. You get through daily sales and correspondence by talking to it — looking things up, drafting and sending all happen inside one chat screen.
What it does
- Instructing work through conversation, with history saved for review
- Sales research via web search and business listings
- Drafting and sending email through Gmail
- Searching accounts and sending DMs through X (formerly Twitter)
- Remembering exchanges and carrying them into the next session
Under the hood
- A web application built with Next.js
- OAuth integration with Gmail and X (tokens stored encrypted)
- An AI agent controlling which tools get called
How it goes
- 01
Where your hours actually go
"We want to do something with AI" is a fine starting point. We map out together which tasks eat how many hours a week.
- 02
Decide what to automate
We separate what AI handles badly from what a person should still check. If something should not be handed over, we tell you why.
- 03
Build small, then test
We run it on one part of the work first and confirm the hours really drop before widening the scope.
- 04
Rollout and adoption
A tool nobody uses is worthless, so we handle the walkthrough and tune the accuracy after it goes live.
About the data you hand over
Credentials created by service integrations are stored encrypted. Before rollout we explain in writing which data goes where. If some data cannot leave your organisation, we establish that boundary first and design around it.
"Can this task be automated?" is a perfectly good question on its own. If the answer is no, we will tell you why.
Talk to us