A Guide for Chatbot Product Managers (Part 1): Evolution

[baoshi.rao]

Since late 2017, I’ve been engaged in chatbot product work and deeply recognize it as a sunrise industry on the fast track. Unlike traditional button-based products, chatbots employ conversational interactions enhanced by AI, which imposes higher demands on product managers. Here, I aim to create a series on chatbot product management—both to help newcomers get started and to facilitate my own systematic growth. This series represents personal perspectives and is humbly titled 'A Guide' rather than a definitive 'Manual.'

This is the first installment of an eight-part series, initially developed as training material for new colleagues. The planned topics include:

In 2016, Google’s AlphaGo 'unexpectedly' defeated world champion Lee Sedol in a five-game match, sparking an AI frenzy. Suddenly, everything was AI, and tech giants raced to invest in the field. Amid this wave, chatbots emerged as a focal point, gaining traction across diverse commercial applications.

Chatbots, as the name suggests, are robots providing conversational services. With over two decades of history, they’ve evolved into a diverse market. This first article explores their origins and development.

Chatbot prototypes appeared over 20 years ago. In 1997, the clumsy 'Clippy' debuted in Office documents. However, users disliked the intrusive paperclip offering unhelpful suggestions, leading to its retirement in 2007.

By 2011, Steve Jobs unveiled the iPhone 4S, Apple’s most successful product at the time, featuring Siri. Initially well-received, Siri became a playful novelty during the mobile internet boom, though its progress over eight years fell short of expectations.

In 2013, Microsoft launched '20 Questions,' a mind-reading game where users imagined a character (or music, etc.), and the system guessed it within 20 yes/no questions. While not a true chatbot due to its button-based interaction, it represented an early experiment in conversational gameplay and became a hit.

A year later, Amazon’s Echo marked a milestone—a highly successful chatbot product. As a smart speaker, it sold over 20 million units in 2017 alone, sparking a trend followed by Google (Google Home, 2016), Alibaba (Tmall Genie, 2017), and Baidu (Xiaodu Speaker, 2018).

In recent years, chatbot development has accelerated exponentially, with new products emerging monthly. Yet, despite explosive growth, quality improvements lag, leaving many experiences immature.

Here’s a brief overview of notable chatbots:

• Cortana: Microsoft’s personal assistant for scheduling, reminders, and emails, built into Windows 10. Despite high expectations, user frustration is evident, with search results prominently featuring 'how to disable Cortana.'
• Xiaoice: Another Microsoft creation, positioned as an entertainment-focused chatbot. With an average of 23 dialogue turns per session, it excels in casual chat but lacks long-term user retention.
• Siri: Apple’s voice assistant, similar to Cortana but stagnant in capability over nine years, limited to basic tasks like setting alarms. Its conversational skills pale next to Xiaoice.
• AliMe: Alibaba’s customer-service bot in Taobao, specializing in order inquiries. More business-oriented, it offers relatively mature UX and includes variants for parenting and casual chat.
• Tmall Genie: Alibaba’s voice-only smart speaker, requiring wake words (e.g., 'Tmall Genie, play a Jay Chou song'). While practical for weather or music, it has minor flaws. The newer CC model adds a screen.
• Xiaodu: Baidu’s smart speaker, a direct competitor to Tmall Genie, with slightly better UX.
• Xiao Ai: Xiaomi’s smart home hub, strong in IoT integration but weaker in wake-word accuracy.
• Wheat: Deployed in China Merchants Bank’s app, this customer-service bot trails AliMe in UX.

This series isn’t a shortcut to becoming a senior chatbot PM. If it helps you grasp the basics, I’ll be gratified. Real growth comes from independent thinking and hands-on practice, so each article includes exercises to spur reflection.

Next up: A Guide for Chatbot Product Managers (Part 2): Classification.