GLOSSARY
Foundation Model
A large general-purpose model trained on broad data, designed to be adapted to many downstream tasks rather than one single job.
A foundation model is the big, general model underneath an ecosystem: GPT-4, Claude, Llama, Stable Diffusion. Trained on broad data at scale, it acquires general capabilities that are then adapted — through prompting, fine-tuning or product layers — into the specific tools people use.
The term captures the economics of modern AI: a few expensive foundation models, many cheap products built on top. It also explains market structure — labs race to build the best foundations, while this directory's tools compete on how well they adapt them to real jobs.