Synthetic Memory

Synthetic Memory is what turns a generic AI into a specific expert or customer. It allows your synthetic personas to have a unique background, specific knowledge, and a history. Instead of starting from scratch every time, a persona with memory understands your business context, remembers past interactions, and responds based on the specific information you have taught it. It ensures your research is grounded in your reality rather than general LLM data.

SYMAR offers two distinct types of memory to suit different research needs. Long-Term Memory allows you to upload your own files to create a private knowledge base, perfect for teaching personas about your products, brand guidelines, or past research reports. Short-Term Memory enables persona to search for and react to breaking news, current competitor pricing, or cultural trends happening right now.

Adding memory significantly increases the quality and relevance of your insights. Without it, the personas rely on the LLM training data. (Which is OK of course, but…). With memory, the persona can act like a loyal customer who knows your product history, or an employee who understands your internal policies. It allows you to simulate complex scenarios where the participant needs to know specific details to give a meaningful answer.

Yes, you have complete control. You create specific “Memory Banks” filled with the documents and data you choose. You then assign these banks to specific personas. This allows you to create different groups with different levels of knowledge, such as testing how a well-informed expert reacts to a new feature compared to a complete novice who has no memory of your brand.