Nineteen case studies, and what each one took

A governed AI platform for investment firms, built on Claude. A firm's own people author, test and deploy agents, skills and workflows, next to the trading dashboards they work in all day, all under a finance-grade reliability bar.

Four years, six products, one on-chain treasury and lending company: marketing site, billing dashboard, block explorer, Treasury Platform. We handed the estate back AI-first, so self-sufficient its non-technical CEO ships his own changes.

A three-sided regulated investment platform: investors, advisors, back office. We rebuilt it page by page while it ran, replacing the component library and the framework underneath with no downtime. Now multi-tenant and white-labelled.

AI due diligence for emerging-market energy finance. We built the platform that assembles a project's financial, regulatory, technical and ESG data, then reads the data room the way a credit team would, built to answer in a day, not weeks.

Founders who knew investments and had never touched blockchain, taken from an AI prototype to a production ownership platform.

US teledermatology platform under HIPAA. We rebuilt the entire consultation funnel, questionnaire, photos, consent, payments and the doctor dashboard, and moved the business from app-first to web-first until most of its traffic arrives in the browser.

On-premise AI video surveillance for industrial operations, no cloud dependency. We built the operator interface: live view, investigation, alerting, subject databases and configuration, where alerting rules are written in plain language.

An early generative AI platform scaled past one product. A single text-to-avatar generator grew into a suite of services, an image studio, a Canva app and the marketing surfaces around them, generating avatars, voice and dubbing at catalogue scale.

An AI image editor for people who cannot edit images, spun out of Synthesys into its own product. We shaped and built it capability by capability: prompt-to-image, masks, brush edits, product restaging. Live, and launched on Product Hunt.

An expert-level exchange on Cardano, and a client who needed a team more than they needed code. We supplied four full-stack blockchain engineers inside their core product from day one, ran the hiring and the management ourselves, and they shipped an order-book exchange to mainnet on the foundation that team laid.

An on-chain vault platform for fund managers and investors: marketplace discovery, portfolio dashboards, vault creation, compliance and multi-wallet flows, with RWA treasuries alongside DeFi yield strategies.

A live lesson-authoring platform rescued end to end. We inherited a five-year-old codebase where nothing could be added safely, mapped it by hand, rebuilt it, delivered the redesign it had been blocking, then shipped new lesson and activity types on the result.

Corporate e-learning that behaves like software. We built the quiz engine, a timed-question video player and the course builder with author and learner roles, plus a path where the client authors custom activities with AI.

Three separate e-learning builds consolidated into one closed course platform, after first cutting a hundred-frame Figma design down to an MVP that could actually ship. The logical closure of the EduLib and DGE Learning story: the lesson authoring and the training modules finally living in one product. Still in active use.

Real-time mission control for aerospace simulation runs. We built the frontend from nothing, a general dashboard, per-simulation pages and a reusable widget set, streaming live telemetry over WebSockets with rendering hand-tuned for a once-per-second tick.

E-commerce profitability analytics, white-labelled: one true profit number out of scattered store data. We took over a part-built dashboard, made it work, then extended it with marketing attribution, plain-language analysis of the numbers and LTV modeling.

A fan-funded crowdfunding marketplace where backers share the upside. We built the entire platform end to end for a solo founder: the feed, wallet-based payments on Stripe, creator analytics, a live ledger and the social layer.

An internal meeting-to-work layer we built in house: CallNotetaker for recording and transcripts, automations and MCP workflows that turn conversation into tasks, in daily use on our own calls.

The finance picture we built for ourselves: incoming and outgoing payments, salaries, invoices, recurring costs and subscriptions in one place, with a dashboard on monthly profitability. Built around our own processes, so we can keep changing it as the agency does.