TestSynthia February 2026 Release: Smarter Personas, Better Targeting, Persistent Memory

We've shipped three major improvements to TestSynthia that make simulated market research more accurate, faster to set up, and more realistic than ever.

The headline features: Custom User Personas that enhance our base AI personas with your specific targeting criteria, Copy Demographics from existing simulations to speed up panel setup, and Persistent Memory for all AI personas so they remember previous interactions and build knowledge over time.

These aren't incremental improvements—they fundamentally change how the platform works. Here's what each feature does and why it matters for your research.

Previously, our AI personas responded based on their demographic profiles—age, occupation, income, interests. Accurate, but generic. The new Custom User Persona field lets you inject specific targeting context that blends with the base persona.

How it works: When you describe your target buyer (e.g., 'A busy working parent aged 30-45 who values time-saving solutions'), our system generates a 'seed template' that places each AI persona into that specific situation. The persona keeps their background and personality, but now they're experiencing your target context.

The result is more relevant feedback. A 'Tech Worker' persona with the 'busy parent' seed responds differently than the same persona without that context. They're the same person demographically, but their current life situation shapes their opinions. It's the difference between 'Would a tech worker buy this?' and 'Would a tech worker who's also a time-starved parent buy this?'

Use this when you have specific psychographic or situational targeting beyond demographics. The more context you provide, the more nuanced the responses.

Setting up demographic filters—age ranges, locations, occupations, income brackets—takes time. If you've already run a simulation with good targeting, you shouldn't have to rebuild those filters from scratch.

The new 'Copy from Existing Simulation' feature lets you clone demographics (and user personas) from any previous simulation in your account. One click copies: age range, gender selection, geographic filters, occupation targeting, education level, income brackets, and your custom user persona.

This is especially useful for: Running follow-up research on the same segment, testing different value props with identical audiences, and team collaboration where one person sets up targeting others reuse.

Access it in Step 3 (Panel Filters) under the 'My Simulations' tab. Select any completed simulation and copy its entire demographic configuration instantly.

Here's where it gets interesting. Every AI persona in our database now has persistent memory. They remember previous simulations they've participated in, the responses they gave, and the context of those studies.

What this means practically: When a persona responds to your survey, they're not answering in a vacuum. They're answering as someone who's participated in market research before. They might reference previous products they've evaluated, express evolving opinions on pricing, or show awareness of market trends based on past studies.

The memory system uses a three-tier architecture: Recent memories (last few interactions), Reflections (synthesized insights from multiple studies), and Core beliefs (stable preferences that persist over time). This creates realistic continuity—a persona who's seen five SaaS pricing pages has different reference points than one who's never evaluated software.

For researchers, this means more authentic responses. The personas develop 'experience' over time. They get better at articulating objections, more nuanced in their preferences, and more consistent in their decision-making style. It's not just demographic accuracy—it's behavioral realism.

The real power comes from combining all three features. Here's the new research workflow:

Start with Copy Demographics to instantly clone targeting from a previous simulation. Add a Custom User Persona to narrow the context further. Run your simulation. The AI personas respond using their Persistent Memory of previous research, filtered through your specific targeting criteria.

The result is feedback that feels surprisingly human—because it's coming from personas with history, context, and specific life situations. They're not generic AI responses. They're the simulated opinions of people who match your exact target criteria, informed by experience, and shaped by the specific situation you described.

Early users report 40% more actionable insights from the same number of responses. The feedback isn't just directionally accurate—it's specific enough to guide feature decisions, messaging changes, and pricing strategy.

These features are live now for all users. Custom User Personas and Copy Demographics are available in the simulation creation flow. Persistent Memory is active across all AI personas in the database.

We're already working on the next wave of improvements: Cross-simulation analysis to track how persona opinions evolve over time, Exportable persona profiles for use in other tools, and Team-shared custom personas for consistent targeting across your organization.

As always, the goal is making market research fast enough that you'll actually do it, accurate enough that you can trust the results, and actionable enough that it drives better decisions. This release gets us significantly closer to that goal.

Try the new features on your next simulation. We'd love to hear what you learn.