4 tips for great conversational AI product management

Conversational AI teams and channels are growing rapidly, which means the demand for product managers in these areas is increasing too. As with any new product area, conversational AI product management comes with its own set of challenges.

We couldn’t think of a better person to chat about this with than the exceptional Brian Smith, who is currently Senior Product Manager, Generative AI Experiences at Intuit. (When we originally published this article in October of 2022, Brian was Conversational AI Design Product Team Manager at Intuit.)

Here’s Brian’s best advice on how to take conversational AI project management to the next level.

1. Challenge designers and developers through rapid experimentation

The first thing every product manager should do is get the product in front of customers for real-time feedback. And they should continue to do so constantly.

Design should be seen multiple times during the design process, not just during user testing. The micro-feedback received from customers during this process can be used to inform designers and developers for more experimentation before moving on to the next subset of customers.

2. Break bots. Then break them again.

Typically, the goal is to put out MVP-ready work. But that status quo approach is safe and rarely enjoyable for the customer.

A product manager needs to first push their team to roll out features that can and will break the existing bot. In this way, teams will learn from what breaks and what works. That is the only way to keep the team creative and the customers delighted. As a result, these welcomed failures will lead to a far better end experience.

3. Communicate wins to the larger team

Good product managers are grounded in the day-to-day and focused tactically on the team’s work. Great product managers also make sure to spread the word widely.

This means celebrating quick wins across teams, showing prototypes to stakeholders, and even communicating ROI to leadership.

4. Paint the journey with horizons

Create a timeline for the entire conversational AI journey. Here, product managers map out capabilities or experiences within the assistant.

Use Brian's horizons framework (below) to create a single snapshot of the customer journey.

Plot out the three horizons on a timeline (informational, transactional, and conversational) and decide with your team what core functions the experience should be able to do at each using customer insights. Then, assign timelines to each horizon.

For example:

  • Informational: answer bank account FAQs (we are here now) (Q3 2022)
  • Transactional: help a user transfer funds (Q1 2023)
  • Conversational: proactively prompt a user to transfer funds based on spending trends (Q4 2023)

Finally, write down and revisit the scope of the digital assistant. This helps everyone design with the user goal in mind and makes any net new requests from other teams an easy decision. Brian says this is very helpful for his teams at Intuit as it sets everyone—leadership, design, developers, data science—on the same page.

Product managers need to think of conversational AI experiences as products and not just features. Thinking ahead and pushing the boundaries of experiences are at the core of the product management role, and these tips can be used as a guide to efficiently carry out that role.

Editor's note: This was originally published in October 2022 and updated in June 2023.


Breaking and rebuilding Nike's chatbot

No items found.