头雁
头雁|11月 21, 2026 12:31
How to Develop Personal Agent Products and Meet Needs? Mona Sarantakos, Product Manager at Muse, revealed some of their methods in her first podcast interview, which is worth watching and should inspire teams developing similar products. Especially how to make users understand that a personal agent is not just another chatbot, and how to discover user usage scenarios (using Agent Swarms for automated data collection). Understanding Personal Agent Needs: She repeatedly stated that their internal motto is "Trust is a must": Personal assistants can access emails, calendars, credit cards, and even family information; a single major mistake could mean losing a user forever. Therefore, they deliberately delayed launch by seven months, spending a significant amount of time on reliability, security, and red-team testing. The Real Needs of Ordinary People Differed Significantly from the Team's Expectations: Mona originally predicted three main scenarios: - Personal Productivity - Health Management - Personal Financial Management However, shopping unexpectedly became one of the most important scenarios during internal testing and among external users. She herself said, "I never expected this." How they "understand" the needs of ordinary people: - Extensive internal trials + external Alpha users: Gathering real feedback throughout the summer. - Agent Swarms: Because Muse's dialogue is a multi-turn, long-duration task, traditional dashboards are incomprehensible. They send out swarms of agents to analyze real-world dialogue data, bringing back specific insights (e.g., "10 people are stuck trying to buy something on a certain website"). Mona requires every PM to review these reports, because they build empathy more effectively than numbers. - Dogfooding: The engineering lead's first task every morning is to send out a bunch of shopping-related bug reports. The biggest challenge: Ordinary people simply don't know what it can do. - This is what she considers the biggest problem. Many people's reaction is, "I already have a chatbot, why would I install another one?" - She gave a vivid example: chatting with parents of her child's classmates, everyone had seen the ads, but no one really understood how to use it. One heavy user, a mother, directly handed over her phone to show everyone how she usually uses it, and two people downloaded it on the spot. Conclusion: No amount of advertising can replace seeing a user see firsthand how someone similar to themselves uses the product. Word-of-mouth demonstrations are more effective than any feature introduction. The core methodology of the Muse team in understanding the needs of ordinary people is—first, establish absolute trust; then, use real-world usage data (especially Agent Swarms) to discover unexpected scenarios; and finally, overcome the cognitive barrier by "seeing how others like themselves use it." Watch here: https://www.(youtube.com)/watch?v=WY0cdD4gNoQ
+4
Mentioned
Share To

Timeline

HotFlash

APP

X

Telegram

Facebook

Reddit

CopyLink

Hot Reads