Master Plan Part 1

This blog explores the assumptions, approach, and open questions behind our plan to make on-device continual learning accessible. Most AI systems stop learning once deployed—we believe they shouldn’t. Our approach focuses on building machines that improve with use—locally, privately, and personally. In short, the master plan is:

• Step 1: Validate the core ideas using concise, targeted demos built on existing infrastructure to test the hypothesis.
• Step 2: Develop an SDK that delivers immediate utility in today’s market, while also serving as a forward-compatible foundation for continual learning.
• Step 3: Create a compelling flagship application, or collaborate as a design partner with developers building such an app.