Human authority
AI should increase human capability without unnecessarily surrendering human authority. Consequential decisions and physical actions require explicit responsibility.
How we work
Atkinson’s values are designed for a company working across institutional AI, sovereign infrastructure, intelligent agents, Physical AI, education and creative production.

These principles are meant to shape architecture, customer engagements, product development, training, creative production and public communication.
AI should increase human capability without unnecessarily surrendering human authority. Consequential decisions and physical actions require explicit responsibility.
Say what is available, demonstrated, proposed, developing or research-oriented. Do not turn ambition into an unsupported claim.
Organizations should understand and retain meaningful control over their data, compute, models, authoritative knowledge, context and operational authority.
Research, intellectual property, records, decisions and accumulated expertise deserve protection and continuity.
An intelligent system should understand the institution, jurisdiction, terminology, culture, language and real environment it is intended to serve.
Technology earns its place by helping people understand, decide, create, learn or operate more effectively—not by being technically impressive in isolation.
Training and enablement are part of implementation. People need judgment, not only access to tools.
Creative work and technical work both benefit from craft, experimentation, rights stewardship, review and purposeful human direction.
AFA’s Canadian identity is expressed through responsibility to Canadian institutions, language, law, public-interest expectations and intellectual property—not through technological isolation.
Commercial, open-source, Canadian, American, European and other global technologies can all be appropriate when they meet the workload.
The organization should determine what sources are authoritative, what context matters, what is sensitive and what actions remain human-controlled.
AFA starts with people, responsibilities, information and desired outcomes.
Start with a Workshop