The word "preventive" shows up a lot in how you describe Tamarynde. What made you decide to position this as prevention rather than treatment?

I think we've normalized waiting until people are struggling before we pay attention to their mental wellbeing. With physical health, we exercise, eat well, and try to stay healthy before something goes wrong. I think we should approach mental wellbeing the same way.

There is a crowded field of AI companion apps right now. What did you see that those products were getting wrong?

I don't think they were necessarily getting things wrong but that they were solving a different problem. A lot of AI companions are designed around engagement, entertainment, friendship, or personality. Others focus on structured mental health exercises. What I felt was missing was a product that was comfortable with listening before trying to solve the problem. When people are overwhelmed, they often don't need advice immediately. What they need is space to think out loud and make sense of what's happening. Tamarynde was built around that idea.

What was the personal moment that made this feel like something you had to build?

There wasn't a single dramatic moment but a pattern I kept noticing. Some of the most helpful conversations I've had weren't conversations where someone solved my problem. They were conversations where someone listened well enough that I eventually found my own answer. The reality is that those people aren't always available when you need them and we are scared to open up to people about our intimate problems. As AI became more conversational, I started wondering whether it could create a space where people could think out loud without feeling judged and help them process what's on their minds. 

Tamarynde is voice first. Most AI emotional support tools are text based. Why voice?

We chose voice because that's how people naturally process things. When you're overwhelmed, opening a text box and typing about how you feel isn't always natural. People talk through problems. They vent, pause, change their minds halfway through a sentence and voice captures all of that. We wanted Tamarynde to feel less like messaging a chatbot and more like having a conversation.

Walk me through what a session with Tamarynde actually looks like.

Let's say you've had a rough day, frustrated about work, worried about a relationship, carrying around a hundred thoughts and don't know where to start.

You talk and Tamarynde listens. It reflects what it's hearing, asks thoughtful follow up questions, and helps you explore your feelings. The goal isn't to rush toward solutions. It's to create enough clarity that you can better understand what's actually bothering you and the next steps to take.

How does Tamarynde know when to keep listening versus when to gently redirect?

One thing we've learned is that people need understanding before guidance. If someone feels unheard, advice usually doesn't land. So Tamarynde starts by listening, it looks for emotional cues and questions that haven't really been answered yet. This balance is something we're constantly refining.

You use the phrase "active listening" to describe what Tamarynde does. What makes it active rather than just responsive?

A responsive system reacts to what you say but an active listener helps you understand what you're saying. Active listening means reflecting emotions, identifying patterns, asking questions, and encouraging deeper exploration instead of jumping straight to conclusions. The conversation becomes less about the AI talking and more about helping the user hear themselves.

Where is the line for Tamarynde? If a user says something that suggests they are in real distress or danger, what does the product do?

We're very clear about what Tamarynde is and what it isn't. It's not a replacement for professional care. If someone appears to be in immediate danger or experiencing a crisis, the priority becomes directing them towards emergency resources. AI has limitations, and I think it's important to be honest about those limitations.

One of the biggest criticisms of AI companion apps is that they can create emotional dependency. How do you think about that risk?

I think it's a valid concern because human relationships should always come first. We don't want Tamarynde to become someone's entire support system. Success to us isn't measured by how much time someone spends talking to Tamarynde. It is measured by whether those conversations help them handle the things they're already dealing with a little better.

There is an ongoing debate about whether simulating empathy is the same as providing it. What is your honest answer?

My honest answer is that AI doesn't experience empathy the way humans do. But I also think that question sometimes misses what matters most. If someone walks away from a conversation feeling understood, that outcome is real despite the mechanism used. I don't think AI should replace human connection and I don't think it can. What it can do is make support more accessible and create space for reflection when human support isn't immediately available.

Data privacy is particularly sensitive in this space because users are sharing things they might not even tell a close friend. How does Tamarynde handle that responsibility?

Trust is everything in a product like this because people are sharing experiences that are deeply personal. That means privacy can't be an afterthought. We believe users should know what information is collected, why it's collected, and how it's protected. If people don't trust the platform, none of the rest matters.

Who is the Tamarynde user you are most trying to reach?

I often picture someone sitting alone after a long day. Maybe they're in their car before going inside, maybe they're lying awake at night or maybe they're replaying a conversation they wish had gone differently. They're not necessarily in crisis but just carrying something and don't have a place to put it. That's the person I think about when we're building Tamarynde.

Is Tamarynde funded or still in an early stage? And what does the business model look like?

We're still in the early stage. Right now the focus is making sure we're solving a real problem. Long term, we see a subscription model making the most sense, but at this stage we're focused on creating a product people want to come back to because it helps them not because we've optimized a pricing page.

The mental wellness market is enormous in Western markets, but access to emotional support tools is arguably even more urgent in places like Africa. Is Tamarynde thinking about those markets at all?

Absolutely, I grew up seeing how access can shape outcomes. In many places, professional mental health support is either difficult to access, prohibitively expensive or heavily stigmatized. AI isn't a replacement for therapists or mental healthcare systems. But it can help lower barriers to support. One of the things that excites me most is the possibility of making emotional support more accessible to people who might otherwise have nowhere to turn.

If Tamarynde works exactly the way you intend it to, what does the life of a regular user look like a year from now?

A year from now, I don't want people to be talking about how much they use Tamarynde.

I want them to be talking about how they recognize their emotions earlier. I want them to have difficult conversations with more confidence and navigate challenges with more clarity. If we do our job well, Tamarynde becomes less about the AI and more about helping people understand themselves better.