Case studies
Systems we build and run.
Every one runs in production today. Read the problem it solves, how it works and the numbers behind it.
An AI platform that turns client meetings into finished reports
How we built and run FinPlan, an application that records a financial adviser's client meetings, transcribes them and drafts the file note or advice document for the adviser to review.
A meeting bot that follows the adviser's calendar
How we built and run calendar-connected meeting capture that joins every online client meeting, reschedules itself when meetings move and leaves when the call is over.
Speaker-labelled transcripts from client meetings
How we built and run a transcription pipeline that separates each voice in a client meeting and names the speakers from the calendar invite.
Readable meeting transcripts that keep the detail
How we built and run a transcript clean-up stage that adds punctuation, paragraphs and corrections to raw speech-to-text output, under strict rules against summarising or inventing content.
Advice documents drafted from the adviser's own templates
How we built and run template-driven report generation that turns a meeting transcript into a file note, fact find or statement of advice in the structure, tone and format each adviser sets.
From draft report to the client's inbox
How we built and run the editing, export and email step that takes an AI-drafted advice document through adviser review and out to the client as a PDF or Word file.
On-demand AI processing with usage-based plans
How we built and run a processing model where meeting audio waits until it is needed, credits are used only when a transcript succeeds and subscriptions run through Stripe.
Three specialist advice workspaces from one codebase
How we built and run branded workspaces for wealth, insurance and mortgage advisers, each with its own look, language and document library, chosen during a six-step onboarding.
Switching AI providers without changing the product
How we built FinPlan so we can move its speech and language models between OpenAI and Microsoft Azure with a configuration change, leaving the business free to choose where its AI runs.
Testing AI report quality before advisers see it
How we built the admin workspace we use to run report templates against real transcripts and recordings, so we can try prompt and template changes before they reach advisers.
Automated compliance document processing for financial advisers
How we built and run a pipeline that pulls the obligations out of each regulatory update, matches them against a financial advisory firm's client portfolios and produces audit-ready compliance reports.
Natural language data queries for portfolio managers
How we connected our natural language data interface to a wealth management firm's data warehouse and run it, so portfolio managers can ask for client data in plain English.
Where to start
Ready for progress you can measure?
Every build starts with Discovery, a paid audit of how a workflow runs today, done with your team. Then we build the workflow and run it for you, every month from go-live.