Five ways we help your team put AI to real work

Claude Cowork for accounting and finance teams

A workshop where AI does accounting work on the team's own real files, from pulling data off tax invoices through to reconciliation. The team leaves with a new way of working, not just knowledge.

What you get

• Talk with the team first to find the most time-consuming, error-prone work, then pick topics that match it
• Practice with the documents and accounting software the team actually uses
• Prompts and Skills that fit the team's workflow, to keep using afterwards

Fits

Organisations holding client financial data, where security is taught from the first hour rather than as a closing footnote

Product requirements for vibe coding (building software by talking to AI)

Turn an idea into a requirement AI can actually build from. Free tools on the page: a readiness quiz, and an AI that pushes back on your idea and returns a PRD outline.

What you get

• A requirements document AI can build from, ready for Claude Code, Cursor, or Lovable
• An ordered plan with a clear definition of “done” for every piece
• A review of your existing requirements, flagging where AI will misread them before you start

Fits

Solo founders, small teams, and businesses that want an internal tool of their own

AI that reads your bills, on your own machines

A pile of bills becomes an Excel file ready for your accounting system. It runs inside your organisation and the data does not leave. Free to try, 30 bills a month.

What you get

• Get an import-ready Excel file for your accounting system, emailed to you
• Bills are deleted immediately after processing, never stored
• Runs on your own machines, fully PDPA-ready, with no cap on volume

Fits

Accounting teams drowning in daily paperwork, and organisations whose data cannot leave the building

Internal tools built with AI

The internal tool your team has wanted for a long time, from the real problem through to a system in use, handed over with a way to maintain it yourselves.

What you get

• A tool running on your own systems, usable from day one
• Source and a handover guide, so the team can keep changing it without us
• An agreed round of revisions after handover

Fits

Teams that already know what they want but have nobody to build it

deban and AI Studio are tools we built this way and use every day

Standing up an agent fleet

A team of AI agents working for you around the clock: scope design, a review system to check their work, and handover with rules your team controls.

What you get

• A set of agents running on your own systems, on the schedule you set
• A signal that tells you they are still working, instead of silence you have to guess at
• A guide for when they break: what to look at first, and how to fix it

Fits

Teams with daily repeating work where a miss goes unnoticed for too long

The fleet we run ourselves works every day, and the process is written up in the articles