CANIVEL · AN INDEPENDENT AI HOLDING COMPANY
AI research.
Model development.
Company building.
We study how AI behaves, train models for specific outcomes, and build companies around what we learn.
Explore our work ↗A natural analogy for improvement.
Our work
Research. Models. Companies.
dyno labAI safety & interpretability
Investigating model behavior, internal representations, and the effects of interventions.
Models for specific outcomes
Model training, fine-tuning, decision modeling, evaluations, and benchmarks.
SCRATCHPADCOLLECTIVEA new company in development
Building a new AI company. More to share as the work develops.
Our approach
Research informs
what we build.
Our work connects AI safety research, model development, company building, enterprise advisory, and investment.
We study how models make decisions, where they fail, and what changes through training. Those findings guide the models and companies we develop.
Enterprise advisory
AI advisory for
enterprise leaders.
We advise boards and enterprise leaders on AI transformation, from model selection and evaluation to adoption across teams and workflows.
Discuss your AI strategy ↗- Strategy & board advisory
- Models & evaluation
- Enterprise transformation
Selected investments
Backing founders.
Writing
On Substack.
Articles and updates from Danilo Canivel.
Contact