Things we've built
No client case studies yet, and we won't borrow anybody else's. Here's what this team has actually shipped and run in production instead.
Why this page looks like this
Most engineering companies our size put three anonymous case studies on this page with a percentage next to each one. We haven't got any real ones yet, so we're not going to write any.
What we've got is systems this team built, that run, that we can talk about in as much technical detail as you want. If you're a CTO deciding whether we've done the thing you need before, that's more useful than a testimonial anyway.
When we have client results, they'll go here. Anonymised, with the client's permission, with real numbers, and dated. If you ever see a number on this page without a date on it, something has gone wrong.
CodeSentinel
An enterprise platform running 14 specialised LLM agents in parallel.
What it involved
Orchestration across 14 agents with different jobs, keeping them coordinated without the whole thing falling over when one misbehaves. Production, not a demo.
Relevant if
You're past the proof-of-concept stage on an LLM feature and finding out that "working in a notebook" and "working on Tuesday afternoon under load" are different problems.
NexusGraph
A distributed graph database written in Go.
What it involved
Distributed systems from the ground up. Storage, consistency, and the parts where the interesting failures happen.
Relevant if
You have a data problem that isn't really a data problem, it's a systems problem, and your team is strong on the application layer.
mettaCofounder.ai
A production SaaS product, built end to end, including the full Stripe subscription lifecycle.
What it involved
The whole thing. Front end, back end, auth, and the payments work most people underestimate: upgrades, downgrades, failed charges, dunning, cancellations, proration, the webhook edge cases that only show up once real money moves through it.
Relevant if
You're adding subscriptions or usage billing and don't want to discover the edge cases in production.
Loyalisto
A SaaS loyalty platform using Apple and Google Wallet passes, with QR ordering integration.
What it involved
Pass issuance, QR-based ordering, and the wallet integrations that make it work in a customer's pocket. Built end to end and shipped.
Relevant if
You're building a loyalty or wallet-pass product that needs to work in the real world, not just in a demo.
Cross-platform mobile app for a German client
A mobile app delivered solo, from architecture to deployment, with zero post-launch bugs reported.
What it involved
Full ownership of the cross-platform build, the backend cloud functions, the iOS and Android release pipeline, and app store review resolution. One engineer, one shipped product.
Relevant if
You need a mobile product on both platforms without hiring a whole team, and you want it to actually ship.
Computer-use agent
An agent that operates a desktop environment autonomously using OpenAI vision.
What it involved
Beyond API integration: the agent sees the screen, decides the next action, and executes it inside a real desktop environment.
Relevant if
You have a workflow trapped inside desktop software and you want to automate it without rebuilding the whole tool.
Twilio outbound voice agent
An outbound voice agent built on Twilio with ElevenLabs TTS and DTMF keypress handling.
What it involved
Voice calls that sound human, handle keypad input, and integrate with the rest of your stack.
Relevant if
You're exploring voice automation for sales, operations, or customer contact and need it to handle real call flows.
A German engineering firm in Munich
A freelance engagement delivered on time with no post-launch bugs reported.
What it involved
Working to a German client's spec and timeline as an external engineer.
Relevant if
You want evidence the working relationship holds up with a German counterparty and not just the code.
The formal parts
- MIME Laureate 2025
- IBM TechXchange Top 100
- Big Data and AI Engineering degree
- EU hosting by default, data processing agreement ready before you ask
If you want to know whether we've built the thing you need
Half an hour with Ilyes, and he'll tell you straight whether it's something we've taken to production before or whether we'd be learning on your money.