hackers building tools for humans
// who we are
Puffership is a small software studio. We build AI-native systems for security, for publishing and for the classroom. Tools that solve real problems. No fluff, no vaporware.
We design around large language models from the first sketch, rather than bolting one onto a workflow that was already finished.
Built by a security engineer, so the threat model gets written before the code does. Least privilege and supply-chain hygiene come as standard.
Small team, fast cycles. We put working software in front of people early and let what they actually do with it decide the next version.
// what we build
Three projects in three unrelated fields. What they have in common is the hard part: getting useful work out of a language model when you can't take its word for anything.
A scanner that reads public code for supply-chain and CI/CD weaknesses, a whole organisation at a time. Raw pattern matches are mostly noise, so every class of finding gets its own triage prompt. What survives, we report.
A publishing pipeline that runs with nobody reading the output. The writer and the fact-checkers sit on different model families on purpose: ask three siblings of one model to check a claim their brother invented and all three will happily agree. Every article is costed to the cent.
Upload the course material, get an exam back. Students answer on paper, you scan the sheets in, and the handwriting is read and marked without you touching it.
See examigo →// in depth
You've got an amigo in me.
examigo brings the power of AI to the classroom, helping teachers give personalised feedback and students reinforce their knowledge.
// reach out
Questions, ideas, or just want to talk shop? Drop us a line.
hello@puffership.com