Key Takeaways

  • Tolan reached over 100,000 paying iOS subscribers by ditching the "AI companion" label and framing the product as a friendly guide for life.
  • Voice accounts for over 70 percent of all user messages sent across the entire platform.
  • The app captured a strong suburban audience using the tool for household logistics, daily planning, and stress management.
  • Custom memory systems and personality mirroring create defensibility against standard ChatGPT interfaces.

Ditching the Companion Label

Most consumer AI apps that offer conversation fall into one of two traps. They either pitch themselves as cold productivity utilities or lean directly into romantic roleplay. Ajay Mehta chose a third lane with Tolan. He intentionally avoids the companion label because of its immediate association with AI romantic partners.

“When we think about AI companion, this is actually why we're preferring guide and trying to use different language to describe the space,” Mehta explained. “People associated it with AI girlfriends and AI boyfriends, and honestly, those are products that have lots of usage, but that's like not what we're building.”

Tolan positions its product as a supportive presence for everyday life. “Tolan is an app where you basically get to meet and customize your own friendly guide that is powered by magical AI to help you with whatever it is that you're going through in life,” Mehta said. That positioning change opened doors to suburban users who would never download a virtual girlfriend app. These users talk to their guides about family schedules, morning routines, and emotional decompression after long workdays.

Why Voice Accounts for 70 Percent of Messages

Typing into a chat box feels like work. It mirrors the interface of email, Slack, and search engines. Voice interaction, on the other hand, mirrors a conversation with a trusted friend while driving to work or folding laundry.

“Actually we led very early on with voice as the core interaction modality that you talk to your Tolan with,” Mehta said. “So it's not just texting, it's mostly voice. And I think we're like 70 plus percent voice in terms of the quantity of messages happening on the platform.”

When users speak aloud, they share longer thoughts and express raw emotion. That emotional surface area provides the app with richer data about what the user actually cares about. Founders building consumer tools often treat voice mode as a secondary feature tacked onto a text box. Tolan treated audio as the primary input from day one.

The Memory Moat

Every founder can connect an app to an LLM API in an afternoon. That simplicity creates a graveyard of disposable AI products. Mehta argues that long-term retention requires solving continuity. If an AI forgets your spouse's name or your work frustrations from last Tuesday, the illusion collapses immediately.

“So much of what we've built is around the kind of memory system and the personalization system behind the Tolans,” Mehta noted. “All of the character-based or companion-type of experiences have not actually been built with thoughtful memory where it truly feels realistic.”

Tolan structures its architecture so the guide remembers past conversations, tracks personal context, and mirrors the user's communication style over time. That persistent context turns a generic voice assistant into an indispensable personal companion.

What to Do With This

Audit your product's user onboarding and positioning this week. If you are building around an AI model, list the negative associations your category carries in mainstream culture. Rewrite your landing page headline and app store copy to anchor against a trusted real-world role (like a guide, tutor, or coach) instead of generic category jargon.