Maggie Chen
Principal Product Manager
Maggie Chen is a Principal Product Manager at Ada, an AI company focused on customer support solutions. With nine years in product and a background in product design, Maggie spent three years as a designer at Ada before transitioning to product management. She led growth design initiatives and now oversees product strategy for Ada's generative AI platform, which serves customers across various industries requiring sophisticated AI guardrails and customization.
A few months ago, my friend Maggie Chen joined us for a conversation here at CDO School. I invited her because she was living through two experiences that many designers are thinking about right now. She's designing with AI and pivoted to product management.
Her journey from designer to product manager wasn't a grand career plan. It was born from necessity. After three years as a product designer at Ada, the company needed a growth PM and no one else had her context about the product and user journey. So she stepped up.
What she discovered was that the overlap between product design and product management is easily 80%. Both roles require deeply understanding who you're building for and why features exist. But the emotional transition proved harder than the practical one.
She spent her first year dealing with identity crisis and imposter syndrome, going through phases of grieving as she said goodbye to that designer identity. I found it particularly interesting how she described her experience of talking with other PMs, and how it's pretty much the same experience design leaders have while talking to others design leaders.
Here are some highlights from the conversation.
There was something that you talked about a little bit, which was how you felt personally about the transition, coming from a strong identity as designer. I wonder if you could just chat briefly about what that experience was like.
I probably dealt with a mixture of identity crisis and imposter syndrome for the entire first year that I was a PM. Growth PM in some ways was an easier transition, because growth PMs have to think about the end to end user journey. That's how growth works. You think about the funnel, and you think about how to transition someone through phases of the funnel, not unlike a user journey map.
But now this year, I'm taking on a lot more scope that is core product work, a lot more API design, technical design. There's been phases of grieving for me, of saying goodbye to that designer identity a little bit. I think design has such a great community. There's this kind of aura around the identity of being a designer. PMs don't have that kind of identity.
At the design events, everyone is largely talking about life and non work things and how that inspires them. At the PM events, they often feel like group therapy sessions where we're talking about how to deal with leadership that doesn't understand user needs, how to get budget to incorporate more research and discovery, how to get engineers to think about what we're trying to build and how that meets customers needs.
A PM is a bit of a shit umbrella. You're there to just take the heat for everything. The transition for me of being able to be a designer and think really freely and creatively in all these different directions the product could go in to now having to just be accountable for things, often things that are not in my control, that has emotionally been the hardest part of that transition.
I'm curious, from your perspective and your background, what are some of the common misconceptions that folks might be talking about or having about AI for those who are just starting to dip their toes in it?
It's the age old challenge of good design. It's about that delta between user expectations and what the technology can actually do. Sometimes in certain capabilities, people expect far more from AI than what it's actually capable of. And in other use cases, it's the opposite.
One of the things that people are really excited about with AI, particularly in my company and in the industry that we work in, is the ability of generative AI to sound quite a bit like a human. You immediately jump to the conclusion this technology is going to replace my human support agents. But what's missing here is actually how much this technology requires you to be quite explicit about how to do things and why to do things.
I often make the comparison that generative AI is a lot like a particularly bright intern. It can learn really fast, and it can do a lot of things, but the rub is in how you instruct it to do those things. If you misspeak or deliver an instruction that's too vague, or you assume this intern understands how things work at your company, but they don't, they simply will just not get it.
We've had an example where a customer was trying to get their AI agent to ask if a customer was using iOS or desktop before giving troubleshooting instructions. They assumed that if they gave the agent all the support articles, it would naturally know to ask that clarifying question. It didn't. It would just assume based on the customer's question and jump right ahead with instructions.
On the flip side, people struggle to understand how powerful AI can be as a catalyst for human creativity and productivity. At Ada, using AI coding tools like Cursor or VS Code AI is now a requirement for all engineers, and it has literally doubled, tripled, and sometimes quadrupled productivity. My husband recently implemented notifications in an Apple Swift UI app expecting it to take a week. With AI help, it took 15 minutes.
I think at this present moment in time, you still very much need a human in the loop. Until we have neuro-technology that can upload everything we know, there's going to be major limits in how these tools are able to interpret the context they work in.
You're using AI day-to-day, you're playing with it, but you also work for a company that is providing this as a service to customers who are providing it to their customers. What are some of the things that your customers are learning about their customers?
This is really interesting, and it's actually one of the challenges of working at Ada. It's not enough just to give people naked access to a large language model and tell them, put that in front of your end users. There is a wide variety of guardrails on how that GPT behaves that is unique to individual customers, specific industries, specific verticals.
Some companies are happy with standard moderation that pauses conversations when users mention dangerous words. Others—like those working with domestic violence victims or in suicide prevention—need the AI to keep conversations going tactfully, referencing specific policies and directing people toward resources.
The heuristic that I've started to go by now is, if 80% of your customer base is asking for a specific capability and they all want it the same way, that's out of the box, it's not configurable. That remaining 20% they get an API or something programmable they can fully customize to their needs. If 80% of the customer base asks for a specific feature and they are split 50-50 on the way they want it, you give them a toggle. If you have a certain 10% of the customers, 20% of the customers asking for something, it's not always an automatic no. Now you have to think about is the feature they're asking for something that fits into our long term goals as a product company.
What precedes both the product and the market are the customer needs and the pain points. It's really being able to read the details of why people use the product the way they do, where they get slowed down, what isn't working, what are the things that are painful that they're not even complaining about because they're not aware of what the pain point is. Those are the seeds that really unlock innovation.
For designers considering a similar transition, Maggie offers a cautionary question: Are you sure you want to?
She finds that 80 to 90% of the PM role is just making stakeholders feel better through whispering and cajoling. And just like design, not all PM roles look the same. In some companies, a PM is just a project manager delivering what leadership dictates. In her experience, she find that the rewarding PM roles are at companies that truly care about user needs and trust that business value follows.
She also offered some resources that helped her make the transition. To prepare, she recommends developing business analysis skills and reading books like Inspired, Escaping the Build Trap, and Product Management in Practice.
But her biggest piece of advice comes from reflecting on her own journey: she wishes she'd felt more confident in her proposals as a designer. That hesitation to assert her intuitions—thinking "I'm not the PM, I can't make decisions"—may have slowed her progression.
After my chat with Maggie, it struck me how much confidence in your expertise matters, whether you're managing design, products, or AI. It makes sense that when we transition into new roles, that our confidence can be shaken because we're both experts and beginners at the same time.
Confidence in your expertise matters, whether you're managing design, products, or AI.
