Automating Product Design with Multi-Agent AI and the Jobs-to-be-Done

I’ve recently enhanced my project User-Centric Products Design with AI Copilot and I wanted to share what I’ve been up to. If you’re into AI, product development, or just curious about how to make building products easier, this might interest you!

Multi-Agent AI

I recently reworked my entire process by using multi-agents (a group of AIs, each with a specific job to do) using the AutoGen tool along with a method called the Agentic Framework.

The goal was to make each step of the product design process more efficient and accurate. Instead of one AI trying to do everything, now I’ve got multiple AIs working together, each focusing on what it does best.

Solution Generation

Another update is the introduction of structured brainstorming. Once the JTBD insights and needs are gathered, we’re now able to generate product ideas and features that are aligned with user desires. This helps ensure that the solutions we design truly address the real-world challenges users face.

StoryBrand Framework

To top it off, I’ve integrated the StoryBrand framework into the workflow. The StoryBrand framework is a powerful storytelling tool for creating copy that resonates with users. Once the product and features are designed, the framework helps craft messaging that communicates the value of your solution clearly and compellingly to your audience.

Full Automated Product Design Process

1. Start with Your Idea

It all begins with you. You bring in a vision, a rough idea or even a problem you want to solve. It can be something super specific like a feature for an app or something broader like a new product concept.

2. AI Proposes Users and Jobs

Once you input your idea, the AI jumps in and starts thinking about who the potential users might be (we call them Job Performers in Jobs-To-Be-Done terms) and what tasks they’re trying to complete (those are the Jobs). For example, if your idea is an app for freelancers, the AI might suggest users like freelance writers, graphic designers, etc., and jobs like managing projects or invoicing clients.

3. Score the Users with a Structured Framework

After the AI suggests potential users, each one is rated using a simple framework inspired by Bill Aulet’s Disciplined Entrepreneurs. This helps prioritize the most valuable users. Here’s what’s considered:

• Funding Viability: Does the customer have the money to buy your product?
• Sales Force Accessibility: Can you easily reach them directly for feedback?
• Customer Value Proposition: Is there a compelling reason for them to choose your solution?
• Whole Product Delivery: Can you provide a complete solution, either alone or with partners?
• Competitive Landscape: Are strong competitors already serving this market?
• Market Scalability: Can success in this market lead to growth in other areas?

The AI rates each user on a scale from 1 to 5 for each of these factors and gives a total score, helping you focus on the most promising users.

4. Pick Who to Analyze

Now it’s decision time. Based on the AI’s scoring, you choose which users and jobs you want to focus on. These are the users who are going to guide the rest of the process, so you can make sure you’re working on something meaningful.

5. AI Creates Synthetic Users

The system generates synthetic users, which are personas that represent your target users. These aren’t real people, but we are using LLM’s power to act like they were almost real. They’re super useful for understanding your audience without needing to do many interviews from scratch.

6. Run Interviews with AI

Once the synthetic users are created, the AI conducts virtual interviews with them using the JTBD framework. This is where we dive deep into what the users are trying to accomplish, what frustrations they have, and what they need. Think of it like having a conversation with a real user, but much faster and more scalable.

7. Extract Insights

After the interviews, the AI pulls out the key insights. This is where you learn what your users’ biggest pain points are and what features or solutions could help them the most. Instead of relying on guesses or assumptions, you’re working with actual data on user behavior and needs.

8. Brainstorm Features

Now that you’ve got the insights, it’s time to brainstorm. The AI helps you come up with ideas for product features or improvements based on the data it gathered. The best part? The suggestions are all rooted in what your users need, not just random ideas thrown at the wall. This ensures that the features you design are likely to be useful and relevant.

9. Create a Clear Copy with StoryBrand Framework

Finally, we tackle the messaging. Once the product or feature is designed, the AI uses the StoryBrand framework to help you create clear and compelling copy. This framework is all about storytelling, helping you communicate the value of your product in a way that’s easy for users to understand. So, instead of just listing features, you’re telling a story that shows users how your product can solve their problems and improve their lives.

10. Why This Matters

What’s exciting about this new approach is that it streamlines so many aspects of the product design process. Instead of spending weeks or even months interviewing users, brainstorming features, and writing copy, you can now do it much faster and still get high-quality results. Plus, the insights you gather are more grounded in real user behavior, so the features and products you design are more likely to hit the mark.

Another bonus is that this setup is much more collaborative than before. Instead of relying on a single AI or team to handle everything, each part of the process is handled by a different AI agent, each specialized in a particular area. This makes the process more accurate and less prone to errors.

If you’re into using AI for product design or just looking for ways to streamline your workflow, this new process might be worth checking out. You can find the project on GitHub if you want to dive into the details, and I’d love to hear your thoughts or suggestions on how to improve it!

Hope that was helpful! Let me know if you have any questions or want to dig deeper into any part of the process.

References

[1] J. Kalbach, The jobs to be done playbook: align your markets, organization, and strategy around customer needs. New York: Rosenfeld Media, 2020.

[2] A. Ulwick, What Customers Want: Using Outcome-Driven Innovation to Create Breakthrough Products and Services. McGraw-Hill, 2005.

[3] B. Aulet, Disciplined Entrepreneurship: 24 Steps to a Successful Startup. Wiley, 2013.

[4] D. Miller, Building a StoryBrand: Clarify Your Message So Customers Will Listen. HarperCollins Leadership, 2017

[5] M. Cello, Supercharging Product Design: Unleashing GPT and Jobs-to-be-Done for Limitless Innovation, Medium, 2023.

[6] M. Cello, User-Centric Products Design with AI Copilot, Medium, 2023.