Artificial intelligence (AI) isn’t just changing the way we build products — it’s reshaping how product teams themselves operate. From design and development to customer insights and training, AI is finding its way into nearly every corner of the product lifecycle. But what does it really take to build an AI-first product organization?
To dig into that question, I sat down with Echo Point Managing Director Mustafa Kapadia to share how my team at NMI is approaching this transformation. In our conversation, I talk about how we’re weaving AI into our daily workflows, the surprising ways it’s helping us move faster and how it’s shifting the balance between product and engineering roles.
Listen to the full interview here or read the Substack article.
Mustafa Kapadia: Today, we’re talking about building an AI-first product organization, something very dear to my heart. To help us with our topic, we have NMI’s Chief Product Officer, Tiffany Johnson.
Tiffany, to get started, can you tell us more about your product organization? At a high level, what is your product org like? How many people do you have, and how are they using AI right now?
Tiffany Johnson: I have a small but mighty group of 17 product managers and designers on my team. Within the engineering organization, we also have product owners and program managers.
My team is focused on value creation, which involves everything from understanding our competition to finding opportunities to help our customers be more successful. We organize around “value streams,” with each value stream having a product director and an engineering director who really own their product and overall strategy. The product managers and designers on their team are all working in service of their value stream North Star KPI.
Mustafa: What’s your ratio between product managers and engineers?
Tiffany: It varies based on the value stream. Some of the front-end ones might have more of a product-to-engineering ratio, while the backend streams might be more focused on execution and less on strategy. So it depends on the product line, but I would say the average is about one PM to 10 engineers.
Mustafa: That’s a healthy ratio. So, within this large organization, how is your team using AI? And what are they using it for?
Tiffany: We’re using AI for a lot of different things. 2025 was definitely a growth year for AI within the product team. We went from using ChatGPT as an advanced search function to see what competitors were doing to folding it into different areas to help us shorten decision cycles.
What I’m really excited about is what we’re exploring from a design perspective. Our designers are using GreenField AI and tools like V0 and Figma Make. We even have some designers playing in Cursor and other applications for coding. We’re also using tools like Gong to record our account managers’ customer calls. We have an automated integration that feeds highlights of those conversations into Slack so we can go back and listen to clips and see what customers are saying.
Mustafa: What has been the most surprising use case that you’ve come across when it comes to how your team is using AI?
Tiffany: There’s not necessarily a single surprising use case, but there are a lot of really interesting ones. For instance, just last week, we launched a new, fairly technical embedded payment component. To help with training, my team found an AI video tool called Clueso. With it, they were able to create a professional-looking internal training video to teach the sales and account management teams how to talk about this fairly technical component and what its value is.
Mustafa: When did you first notice that AI was something that your team was starting to rely on?
Tiffany: It kind of snuck up on me. When you look back, it wasn’t that long ago that we were just using AI for ChatGPT’s research function. Now it’s part of our everyday cycle. I don’t know that there was one single catalyst, but there was a lot of internal and external pressure. We also started to realize that without AI, we weren’t going to be able to keep up with some of our engineering efforts.
Mustafa: So it looks like it started with engineers using AI more for the coding piece, which then created extra capacity. Is that correct?
Tiffany: It’s certainly part of it. But we also realized that the user feedback loop was taking too long. For instance, we don’t just need feedback from our channel partners, but from merchants and cardholders as well. If we have an idea, how do we build a prototype and get feedback? As with any product organization, cutting down on that decision-making and trying to solve problems and get user feedback faster is important. AI is helping us address those roadblocks.
Mustafa: Your team is using AI, which is fantastic, but it’s also forcing you to rethink the way you organize your teams. So, tell me more about that thought process and where you are in that reorganization mandate.
Tiffany: That’s a great question, and it’s certainly top of mind. You asked me about ratios earlier in our conversation, and I said we’re about one to ten (product managers to developers), depending on the product.
I’m part of a group called “Products That Count,” and the general sentiment among CPOs in our council is that while most organizations are working with a ratio of 1:10 today, by the end of next year, that will evolve towards a ratio of 1:2 product managers to engineers.
Engineers are becoming so more efficient with AI. But what AI doesn’t replace yet is context, decision-making and empathy. It can’t make big decisions. So, the capacity and throughput and role of the product manager versus the engineer will be a dynamic that shifts over the next few years. That ratio is very much top of mind as we think about our 2026 budget and what that team structure looks like.
Mustafa: How are you determining that new ratio?
Tiffany: Anytime you invest in new department growth, there needs to be an investment thesis behind it. For instance, what’s the return? What’s the timeframe? We’re also considering whether we need to hire new people or if there are engineering leaders (or people in other functions) familiar with the product who we can cross-train to make those product decisions.
We’re on a learning curve, and what we’re finding is that some of our front-end teams are realizing 50% or 100% efficiency gains with AI. In contrast, some of our backend teams who are working with that monolithic code base are actually seeing efficiency declines by using AI because it hallucinates or causes issues. That causes them to spend more time troubleshooting than actually developing.
So, we really need to see how AI evolves in that sense before we throw more product people at it. If there are certain value streams that don’t benefit as greatly, we need to keep our ratios consistent. We don’t want to be in a position where we have too much product management and not enough engineering. That’s why I think we’re also beginning to see a blending of roles. For instance, a designer can do more of what product managers used to do and vice versa, so it’s a very interesting place to be.
Mustafa: As you’re going through this transformation, what are the biggest challenges you’re facing right now?
Tiffany: I’d say there are two key challenges. One is figuring out how to support our teams and engineers so they have the white space needed to learn AI and get comfortable with it. They have commitments to customers and a packed roadmap, so how can we help carve out some time for them to learn?The other challenge for me as a product leader is determining what our investment thesis will be in 2026.
Mustafa: You bring up a good point. I think a lot of CPOs are struggling to show how their teams are being more efficient and productive with AI. So I’m curious, how are you approaching that problem? What are you using as an indicator to show that things are moving faster?
Tiffany: Our North Star indicators tend to revolve around revenue, as they should. Enterprise value creation is important. So, what we’ve done for each of our teams is set what leading indicators they’re responsible for. That has made it really easy for us to look at output metrics and see how productive our teams are. But, that said, none of that really matters unless the outcome is influencing that North Star.
So that’s what we’re trying to figure out right now. Engineering has a lot of great metrics on how they’re delivering lines of code faster and more efficiently. Now, product is working on our target OKRs and KPIs for next year. For instance, are we really moving the needle faster? That’s what we’re trying to use to measure the value of AI on our side.
Mustafa: It’s great to know that you’re leading the charge. You are way ahead of the game. A lot of people are just trying to figure out how they should even think about AI. But you’re not just thinking about how your people are using it, but also how it can transform your culture, processes, and generally the way you work.
To listen to Tiffany and Mustafa’s full conversation, you can watch their interview here.