Products
Things you can open and use right now — built here, free, and aimed at a problem someone in Ghana actually has. It grows every cohort.
Confounded — a gym for scientific judgment
AI can now draw a flawless-looking scientific chart in seconds — including when the science behind it is broken. Confounded is a training gym for catching those: we hide a real flaw in a real experiment, let the AI draw the chart, then ask you (and three Claude models) to spot it. Almost nobody can, from the picture alone.
Try it ↗Code ↗Kwan — guidance and counselling for the choice after BECE
Kwan is Twi for "way". Choosing a Senior High School shapes a Ghanaian student's whole trajectory, and families make that call with almost no structured information — and rarely with a counsellor in the room. Kwan is not a school directory: it takes a student and their family through preferences, constraints, programme pathways and trade-offs, then down to the eight school-and-programme choices the CSSPS form actually asks for, on top of a mapped register of SHS and TVET schools across all 16 regions. Guidance here means making the consequences of a choice visible and checkable, never handing out advice: any programme we cannot source from the official register is marked as unconfirmed on screen. Placement still happens through CSSPS.
Try it ↗This space is for what you build
Fellowship builds, hackathon projects, and open-source contributions land here with your name on them.
start something →Services
What we do alongside a partner lab or team: put an AI collaborator on data that is too big to work through by hand, and leave the scientists free to do the science.
Mining a genome-scale CRISPR screen in human T cells
A diaspora↔Ghana team mined a ~2TB Perturb-seq screen of human CD4+ T cells in real time, hunting master-regulator genes — with an AI collaborator handling the pipeline so the scientists could focus on the biology.
with Kwarteng Lab (KNUST, Ghana) · Greenbaum Lab · Paz Polak
built on Anthropic Claude Science AI Workbench
$ Have a dataset nobody has had time for — life sciences especially? If you have the data and the question, we can put a diaspora↔Ghana team and an AI collaborator on it. Talk to us →