Adapters are where business IP starts to live
Adapter-led AI makes domain knowledge portable, governable, and cheaper to operate than repeatedly prompting frontier models.
Opinions Series
Teaching Smaller Models
7 posts
Teaching Smaller Models
Deterministic code is the secret to smaller models
Smaller models become far more capable when code handles the parts of the workflow that should never be guessed.
Teaching Smaller Models
Cutting AI cost without cutting capability
Model upskilling gives businesses a way to reserve expensive models for teaching, exceptions, and genuinely complex work.
Teaching Smaller Models
From prompting to model upskilling
The next operating discipline is teaching smaller models to carry repeatable reasoning inside governed agentic systems.
Teaching Smaller Models
Large models as teachers, smaller models as operators
Frontier models are often best used briefly and deliberately, so smaller models can carry routine work at lower cost.
Teaching Smaller Models
Reducing AI carbon by moving reasoning closer to the work
Smaller adapted models can reduce unnecessary inference overhead when they are used for well-specified, repeatable tasks.
Teaching Smaller Models
Why smaller models need teachers
Smaller models do not need to know everything. They need to be taught the task, the business, and the boundary conditions.