Insights

Research, field notes and results

What we are learning across strategy and engineering engagements — published as we learn it, not sanitised a year later.

Technology
July 6, 20267 min read
Build the evaluation harness before you choose the model
Teams pick a model, build a demo, then try to work out whether it is good enough. Reversing that order changes the economics of an AI programme more than any model choice.
Priya Raghunathan · Partner, Data & AI

Showing 1 article tagged “Evaluation”.

Want this thinking applied to your business?

Everything we publish comes out of live engagements. Tell us what you're facing and we'll tell you what we've seen work.