Structure of a Synthetic Individual

Each individual is made up of four components. Understanding what each one does helps you know where to intervene when you want better results.

• Profile — identity, background, and defining traits. This is the foundation. Everything the individual says is grounded here. If responses feel off, the profile is usually where to look first.
• Reflections — behavioral analysis derived from their responses. As the individual interacts, iMario builds a layer of interpretation on top of raw answers — patterns in how they reason, what they prioritize, and how they tend to respond under pressure. Reflections are what make an individual feel like a person, not a prompt.
• Capabilities — knowledge and skills, which can be extended with Agent Skills. Out of the box, an individual knows what their profile suggests they would know. Capabilities let you go further — injecting domain expertise, product knowledge, or specialized reasoning that makes them more useful for specific tasks.
• Memory — a record of past interactions that updates over time. This is what separates iMario from a stateless chatbot. Memory means the individual remembers what happened last time — and responds accordingly. An individual you've worked with across ten tasks is meaningfully different from one you've just created.