A new category of intelligence.
Learning that begins where training ends.
AI models learn before they are released, and then stay as they are. ACI learns and acts after: at runtime, from your work, for as long as it runs. That makes it a category of its own.
Where training ends, ACI begins.
A model’s learning ends when it is released. ACI’s begins with your first request, and goes on for as long as it runs.
Step 1 of 5: Training
A model learns the world in one long training run, before anyone uses it.
It builds on what came before.
Each result adds to what ACI knows, without disturbing what came before. A new rule applies from the next request, and everything else it learned stays as it was.
Step 1 of 4: The new and the old
It is intelligence, so it generalises.
ACI learns the pattern behind a result, so it handles a case it has never seen, tells a look-alike apart, and acquires skills it tests before use.
Step 1 of 3: New words
The same request in new words: ACI knows it, and answers.
It learns your work, and what it learns is yours.
A model knows what the world wrote down. ACI knows what your work has confirmed, shows where each piece came from, and goes wherever you work.
Step 1 of 4: A decision
A decision your team confirms becomes something ACI knows.
See Duplex on your own work.
Duplex is built on ACI. A pilot puts it on your own work, with the measures agreed before it begins.