Traditional vs. AI-Enabled Product Development

How AI is changing the product development lifecycle

Hey, I’m Colin! I help PMs and business leaders improve their technical skills through real-world case studies. For more, check out my live courses, Technical Foundations and AI Prototyping.

---

Product development is undergoing a fundamental shift. While the core principles of solving customer problems remain unchanged, AI prototyping is revolutionizing how we validate and refine solutions. Let's break down both approaches and see how AI is transforming our workflow.

Traditional Product Development

Product development is a two-step process of exploration and prioritization.

First, your team determines what problems are valuable to both customers and the business. Then, you deliver an effective solution.

Creating a solution is not simple. Once a specific problem has been identified, the team has to:

1. Write a PRD that describes ideal end-state for a solution

2. Explore one or more solutions

3. Create mocks in Figma that show how the solution could look

4. Gather customer feedback on mocks (ideally)

5. Create implementation plans

6. Ship the initial feature and gather additional feedback from customers

7. Continue with additional iterations

This is considered to be a good product development lifecycle, with lots of customer feedback in the loop and clear communication between teams.

Unfortunately, real life doesn’t always work this way.

• Customer feedback changes after customers actually use the product

• Internal teams aren’t as aligned on the solution as you thought

• Exploring multiple solutions is time intensive, so teams often only test one idea

• Iterations with engineering are very expensive, and even design iterations can take a long time

I think Teresa Torres does the best job of explaining the tradeoffs of exploring multiple solutions here:

“Decision-making research tells us when we compare and contrast our options, we make better decisions. As a general rule, I recommend that teams consider more than one solution for their target opportunity.

In reality, we don’t always have time to do this. Compare and contrast when there’s risk in the solution, when the opportunity is a differentiator, or you need to uncover an innovative solution.”

Fortunately, AI-enabled workflows allow us to try more solutions, get feedback faster, and more quickly align internal stakeholders.

Product Development with AI Prototyping

AI-enabled product development starts the same way, with identifying what problems are worth solving. This still requires product managers to have a strong understanding of their customers, the market, and internal stakeholders.

After this, the process changes substantially:

1. Write a PRD that describes ideal end-state for a solution

2. Create 1-3 AI prototypes of solutions

3. Refine your thinking & solutions based on the team’s experience using the solutions

4. Get customer feedback on 1-3 solutions, where the customer interacts directly with the prototype

5. Finalize the solution, using your prototype to communicate scope internally

6. Create implementation plans

7. Ship the initial feature and gather additional feedback from customers

8. Continue with additional iterations

The critical difference with an AI-first approach is that your initial prototype will take you 1-2 hours, and each subsequent iteration will take 15-30 mins. You can explore 3, 5, or 10 solutions in less time than it used to take to design one.

Customers will also give you real feedback based on their actual interactions with the application – not a static design. Many small friction points may appear as users navigate through typical workflows with your new feature prototyped on top.

When it comes to building AI features, this approach is even more powerful. Get feedback from customers as they call an actual LLM and get real responses, rather than mocking what the LLM might return in Figma.

If you want to double check user workflows, add PostHog or another product analytics tool and watch recordings of user interactions with your prototype.

All for free and in less than half a day of total effort.

Example: Teams AI Summary

Let’s imagine you’re a PM working on Microsoft Teams. One of the key challenges customers have brought up is losing track of tasks throughout the day. You feel that a solution to this problem could be valuable enough for users to opt into a higher tier.

To start, you define the opportunity and the characteristics of an effective solution:

• Simple to understand

• Personalized to each user

• Tracks missed messages and important deadlines

• Includes tasks from both live meetings and text messages

• Syncs with Tasks to push items to your list automatically

You review your draft PRD with engineering and design early to gather ideas and align the team to the upcoming project. Once you have clarity on the solution scope, you begin creating prototypes.

You start with the regular Teams interface:

Credit to Malcolm Goldsmith for the MS Team prototype

Then, you add your new view, Daily Summary.

Credit to Malcolm Goldsmith for the MS Team prototype

After you nail your initial prototype, you generate two other iterations. You book calls with customers to collect feedback, or provide them an asynchronous guided experience where you can view their interactions. Based on this feedback, you further refine your prototypes.

Once the user workflow is finalized, design delivers the final workflow based on your design system and engineering kicks off implementation. The prototypes serve as an internal communication tool – engineering asks what specific AI content should be returned, and design bases the overall design on the prototype while blending in existing patterns and interactions.

The feature ships to productions, users provide more feedback, and you continue to iterate.

While AI prototyping is powerful, it doesn't replace fundamental product management principles:

• Don't jump to prototyping before thoroughly understanding the problem space. A beautiful prototype of the wrong solution is still the wrong solution.

• When showing prototypes early, consider creating multiple variants to prevent anchoring on a single approach. The goal is to validate solutions, not commit to the first idea.

• Finally, while prototypes are great for visualizing solutions, your PRD should still cover critical non-UI elements like go-to-market strategy, success metrics, and technical considerations.

Putting It All Together

AI prototyping isn't replacing traditional product development—it's augmenting it. The fundamental work of understanding customers, prioritizing problems, and executing well remains unchanged. What's changing is our ability to validate solutions earlier and communicate more effectively across teams.

This shift enables product teams to:

• Get more value from customer conversations

• Reduce wasted engineering effort

• Build stronger alignment across stakeholders

• Ship with higher confidence

These tools are accessible to any product manager today, regardless of technical background. There's no reason not to incorporate AI prototyping into your workflow—it's becoming as essential as writing PRDs or creating roadmaps.