Home Work
Case studies — or, honestly,
what we've built.
Here's the truthful version: we're a young studio, and we don't have client logos or five-year testimonials yet. What we do have is production software we designed, wrote and run ourselves — on exactly the stack we'd build for you. That's the proof on this page.
The honest version
No client logos. Real software instead.
The same standard as our About page: fake reviews — zero, invented metrics — zero. If we haven't done it, we won't claim it.
A lot of agencies open with a wall of borrowed logos and rounded-up numbers. We'd rather show you the software we live with daily than the clients we don't have yet. Every product below is something we scoped, engineered and now operate ourselves — with the bugs, the uptime and the design decisions all on us. When we recommend a pipeline, a mobile stack or a verification step, it's because ours has been running on it around the clock, not because it read well in a deck. Judge the engineering. That's what you'd be hiring.
Built & running
The software we operate.
Six build stories. One public product, two pieces of craft you can click, and three internal tools we built to run our own operation.
Unbuilt Lab — startup-idea research, live
The problem. Founders guess at what to build instead of reading real demand signals. What we built. Unbuilt Lab discovers startup ideas end to end: it scrapes demand signals across the web, clusters them with NLP, generates candidate ideas with AI, then scores each one across multiple dimensions so the promising ones surface. Subscriptions, entitlements, quotas and billing are all part of the product, not bolted on. The engineering. Python and Flask over PostgreSQL, with Celery workers running the scraping and processing pipeline around the clock, and a web dashboard on top. This is the AI-plus-data pattern behind our automation and research work, running for real and paying its own way. Honest status. Live and running as a web-based SaaS — our flagship public product.
Stack — Python/Flask · Postgres · Celery · web app
Pipeline — scrape → cluster → generate → score
Status — live, public product
The template library — front-end craft, live
The problem. "We do great websites" means nothing without something clickable. What we built. Nineteen complete, animated marketing websites — every one hand-coded here, no themes and no page builders. The engineering. Hand-written HTML, CSS and JavaScript with self-hosted fonts, designed light and dark modes, and load times that keep both Google and visitors happy. Honest status. All nineteen are live and clickable right now, and any of them can be re-forged around a brand in days.
Stack — hand-written HTML/CSS/JS · self-hosted fonts
Proof — 19 live, all animated
This site & its chatbot — the portfolio piece you're on
The problem. An agency site should demonstrate the work, not just describe it. What we built. The site you're reading is hand-coded top to bottom — no page builder anywhere — and in the corner is a self-hosted RAG chatbot, grounded in our own content so it answers from fact and hands off when it should. The engineering. Hand-written front end plus a self-hosted LLM/retrieval stack (Dify) we run ourselves rather than renting a black box. Honest status. Live — this page, and the assistant in the corner of it, are both the proof.
Stack — hand-written front end · self-hosted LLM/RAG (Dify)
Proof — you're using it
Internal tooling
Built to run our own operation.
Not products we sell you — infrastructure we built for ourselves, and the reason we can talk about pipelines, deliverability and outreach from experience rather than theory.
LeadProwl — our lead-data infrastructure
The problem. Stitching together scrapers, enrichment tools and verification services that don't talk to each other burns hours and produces dirty data. What we built. LeadProwl is the internal lead-data platform behind our own pipeline: scraping and data collection, decision-maker enrichment, email verification and outreach, wired as connected stages rather than a pile of subscriptions. The engineering. Python and FastAPI over PostgreSQL, with Celery-backed queues and schedulers doing the heavy, around-the-clock work so a slow enrichment step never blocks the rest. Honest status. Internal infrastructure, in active development — and the system our lead generation service is built on.
Stack — Python · FastAPI · Postgres · Celery queues
Role — runs our own pipeline
NoMailBounce — our verification engine
The problem. Sending to dead or risky addresses quietly wrecks deliverability and sender reputation, and buying verification by the seat gets expensive fast. What we built. NoMailBounce is our own email-verification engine with a clean API — syntax, domain and MX checks, disposable and role detection, and SMTP probing that returns three honest verdicts rather than pretending certainty about catch-all domains. The engineering. A React/Vite front end over a Node API, callable directly or as a service, and embedded as the verification stage inside LeadProwl. Honest status. Internal infrastructure — and the engine behind our bulk email verification service, so clients get our own code rather than a reseller account.
Stack — React/Vite front end · Node API
Role — powers LeadProwl and client verification
Boldarity — our outreach and demo system
The problem. Telling a business with no website that it needs one is abstract; showing them their own site is not. What we built. Boldarity automatically builds a sample website for a prospect that has none, then runs the outreach around it — messaging, an AI-assisted reply flow, and the handover if they say yes. The engineering. The same pattern we sell as automation: scraping and enrichment feeding a generation step, then a messaging pipeline with a human in the loop before anything reaches a real person. Honest status. Internal tooling — the practical test bed for our process automation and site-generation work.
Pattern — scrape → generate → outreach → handover
Role — proves the automation stack on ourselves
What this means for you
You get the same engineering — and you own it.
The person who built and runs these products is the one who builds yours. The scars are already paid for — your project starts on patterns proven in software we operate daily.
We ship the smallest honest slice that proves value, so you learn from real users before over-building — the same way we grew each product above.
Hand-written code, documentation and credentials handed over. No page builders, no black-box subscriptions, no lock-in — the code is yours to inspect and keep.
Want the detail on how each capability maps to a build? The services page breaks it down, and our write-up on mobile app case studies and what actually makes them succeed is the same thinking applied to product work.