Community sentiment intelligence
We read and clustered 1,729 community mentions, then triaged them into 8 product insights and 4 critical alerts in about an hour. Every recommendation carried a named owner and a deadline.
Work
The first list is client work, described accurately but without naming the client. The second is what we built for ourselves, named in full, because you can open it right now.
Client engagements
We withhold client names by default. Sector, scope and outcome are accurate; identifying detail is left out. Ask us and we'll arrange a reference conversation.
We read and clustered 1,729 community mentions, then triaged them into 8 product insights and 4 critical alerts in about an hour. Every recommendation carried a named owner and a deadline.
Someone used to assemble the daily sales report by hand. We rebuilt it as a pipeline that runs overnight and has the report waiting when the team logs in.
We mapped inquiry to sales order to invoice end to end, found the bottlenecks, and handed over a three-phase digitisation roadmap.
A registry that keeps every contract accounted for and surfaces permit obligations before they lapse.
We moved records out of an operational system into a warehouse the business can query directly, so reporting reads from one source instead of five exports.
Our own properties
The studio's main hub.
Published studies with methodology and sources in the open. The flagship study is “AI Finance Department”.
Course hub. Live: “Claude Code Mastery” and “Managerial Finance I+II”, both bilingual ID/EN.
How one AI assistant becomes a squad of agents, written up from practice.
An AI drying-QC simulation, playable in the browser, EN/TH.
Tell us where the work piles up. We'll say whether AI should touch it, and what a first phase would take if it should.