AI
Grounding LLM features in your own data without shipping a hallucination
Retrieval is the easy half. The hard half is deciding what your system does when it is not confident — and measuring that before launch.
Ines Duarte
9 min read
Writing
Notes from inside the work.
What we learned on real engagements, including the decisions we would make differently. One issue a month, no sequences.
6 articles
Performance
What a Lighthouse score does not tell you
A perfect lab score and a slow site are entirely compatible. Here is how we instrument real-user performance instead.
Marcus Ainsley
7 min read
Engineering
The handover is the product
Agency work has a failure mode: excellent software nobody internally can extend. Documentation is not the fix.
Marcus Ainsley
6 min read
Design
Designing for gloves, bad light and no signal
Field software breaks every assumption a desktop dashboard is built on. What we learned shipping to 1,100 engineers.
Yuki Tanaka
8 min read
Product
How we scope fixed-price software honestly
Fixed price only works when the discovery phase is real. Here is the shape of the estimate we actually stand behind.
Sofia Ferreira
7 min read
Engineering
When not to automate a process
Automating a broken process makes it faster and harder to fix. Three tests we apply before writing a pipeline.
Sofia Ferreira
5 min read
Want us to look at your version of this?
A 30-minute call, no deck. We will tell you whether this is a project we should take, and what we would do first.