Software we build and run ourselves.
Two products, both in beta, both ours end to end — the data platform, the game, the hosting, the billing and the pager.
We think a consultancy should be able to point at things it owns. Client work is mostly covered by confidentiality agreements, and a slide deck full of logos proves very little about whether the software works. So this page is about the products we run: one commercial data platform and one game, both built in-house, both maintained by the same people who would work on your project.
Between them they cover the two halves of what we do — a system whose job is to not lose money, and a system whose job is to be interesting to use.
Data Platform · Beta · Invite Only
MochaFlux
A SaaS platform that schedules Amazon reports, downloads them, processes them and loads the results into Snowflake — without anyone writing or babysitting a script.
It handles the unglamorous parts that make report ingestion painful in practice: API throttling, retries, queueing, schema drift, and loads large enough that a naive implementation falls over.
Access is by invitation while we finish the rollout. Tell us about your pipeline if you would like in.
Read about MochaFluxGame · Beta
NanoGenesis
A free browser idle game about growing a self-replicating nanobot swarm until it can enclose a star in a Dyson sphere. Sixty elements, ten star classes, one very long number going up.
It runs entirely client-side, installs as a Progressive Web App, and simulates your time away. It is also an experiment in game design: the systems are ours, and AI turns the design into working code and artwork.
Read about NanoGenesisClient work
Most of what we build for clients is under NDA, so it does not appear here by name. What we can describe is the shape of the work, because it repeats more often than you would expect:
- Reporting that has outgrown its spreadsheet. Data that arrives as CSVs from a marketplace, an ERP or a dozen inboxes, and needs to land in a warehouse on a schedule without a human in the loop.
- A warehouse that costs too much. Snowflake accounts where the bill grew faster than the data — usually a warehouse sizing and clustering problem with a query-pattern problem underneath it.
- An internal tool nobody wants to maintain. A process that lives in one person's head, rebuilt as an application with tests, a schema and documentation.
- A product that needs a second engineer. Existing SaaS codebases that need architecture work, a migration, or simply someone senior to carry a feature through to production.
If one of those sounds familiar, that is usually enough to start a conversation — see services, or just tell us what you're looking at.
How we build
The same practices apply whether the software is ours or yours: boring technology where boring technology is correct, tests around the parts that cost money to get wrong, a written scope with a defined end, and documentation a new engineer can actually follow. Everything we run — including this page — is deployed from version control with no manual steps.
We also use AI as a tool rather than an author, on our own products and on client work. It generates against a design we wrote and runs the review passes for security, performance and documentation; a named engineer signs off every change. The detail is in how we use AI.
Want the same treatment?
Tell us what you're building and we'll tell you honestly whether we're the right people for it.
Start a Conversation