“Pivot your team immediately-this is going to transform education.
This is what Duolingo’s CEO said when he got early access to GPT-4, and tried it out before (almost) anyone else.
He booked a meeting with his Principal PM, and brainstormed the newly shipped feature: a learning experience powered by AI to practice conversation skills.
The team spent months testing this technology with small groups of learners, and one of their challenge was to train the data quickly, build new features and a new subscription tier.
It is innovative product wise, but implementing AI also changed how the team operates at an organizational and process level too.
It involved a change in mindset (what can we solve with this technology), collaborating with AI research scientists, or validating prompts and training the model as a team (and not only technical people).
Duolingo Max’s Lead Engineer mentioned how “it changed our engineering process internally. The features we’ve put together have come out faster than they would have before GPT-4.”.
AI-powered products are popping up everywhere: Figma bought Diagram, Framer announced a new AI feature as well as Superhuman, Canva revealed the new visual worksuite, and the no-code tool Softr allows every builder to integrate AI.
And it seems that to stay competitive on the job market, you need to become a “superhuman” AI product manager. It can trigger two reactions from you: a deep excitement for possibilities or fear that AI will replace your job.
Both are valid, but both can be positive stressors to move you towards curiosity.
Marily Nika, computer scientist and AI Product Manager at Meta is deeply optimistic and believes that AI is not replacing our job but enhancing it. She encourages every Product Managers out there to read a lot about it, learn how to train small models, and learn to work with new team members such as researchers.
If we want to go deeper into the job and team changes, here are some key questions we’ll cover:
- How to build AI into your product?
- How can you level up to become an AI Product Manager?
- What does an AI product team looks like?
- Find AI Jobs, including AI Product Manager, AI Researchers, Data Scientists and AI Product Designer Roles
How to build AI into your product?
Don’t do AI for the sake of AI, but the mindset that tend to shift for AI Product Manager is—identifying opportunities behind your product’s data that can be harnessed with AI, while solving a user pain point.
Any product where you can get data behind the behavior with users can be improved with AI, says AI PM at Meta. “If you're working on any product, you can always sprinkle in a smarter feature. So you can make it more secure, you can personalize it, you can enhance it with fraud detection, you can make it more ethical. If it's healthcare, you can make it faster, you can make it more accurate. If it's shopping, you can create better accommodations. Basically anything where you can get data behind the behavior with users can be improved with AI.”
Duolingo Max’s release story is a perfect example. For them, it required:
- A “data” mindset change: “Okay, we have all this data lying around, what should I do with it?”. It’s actually the opposite from what product managers usually do which starting from a user problem - now they start from “what can I solve with this technology” or “which business opportunities can I vet from this data and using AI?”
- A skill upgrade: Getting curious, and learning how to build, train a machine learning model, and when to validate prompts. It definitely requires a tech and data curiosity, and the ability to lead such a project. You will face new problems such as, how much data do we need? Where do we find it? and can we actually use it? — that’s where team work is key and takes us to the last point.
A team work upgrade: you will work with data scientists to collect & sort data, AI researchers to train the model, and engineers to build it.
Embrace uncertainty: It can take time to train enough data, and it’s even harder to know if a new product feature will work out. Duolingo quickly decided to cover the cost, launching it as a paid feature, but they also benefitted from a solid user base as well as an important existing set of data (they tested machine learning experiments for a long time). These are not given if you’re in a smaller team, and something to consider.
How can you level up to become an AI Product Manager?
Many believe that AI will be baked into every product we use, and thus PMs will become AI Product Manager. And contrary to the belief that it will replace your job, it might enhance it.
Marily, AI Product Manager at Meta shares in Lenny's podcast that non-technical PMs should not be intimated: “I'd like to see people being less overwhelmed, less intimidated, less afraid to start learning how to code, how to train a little model on their own.”
And invest in themselves to upgrade your Product Manager’s skills. Here are a few ways:
- Read a lot on the topic, to navigate the trends and stay up to date. You can learn the basics with Introduction to AI by Stanford, and then read Lenny’s Newsletter, as well as Alpha Signal (a weekly summary of top research papers released) or Ben Bite’s AI newsletter.
- Learn how to train your own models. If you’re a non-tech, you can try to bring your own machine learning idea to life with Lobe.ai.
- Learn how to become an AI product manager with Marily's course: AI Product Management for aspiring and current PMs.
- Be curious on working alongside new kind of team members including AI researchers, which can be a change!
What does an AI product team looks like?
As an AI product team, one of your main objectives will be to build models, implement those into your products and make your features successful business-wise. The main bulk of AI-features is the data fed into it, and to provide great data, train it properly and make sure the gathering of data is ethical, then your team will certainly be more research-focused and data driven.
Even though the technology is always evolving and the resources required dependent on your company’s and team’s size, teams at Meta, Duolingo, Microsoft tend the hire the following roles:
- Generalist PM who helps their team and their company build and ship the right product.
- AI PM who helps their team or company solve the right problem (like at Grindr, or Runway)
- AI Research Scientists to lead research and development and train machine learning models (like at Duolingo, or Loom)
- Data Scientists to collect, sort and train data and integrate models in the code base (like at Voodoo)
- Product Designers to identify how you solve a user's need with a clear user experience (like at Genies to work on AI avatars)
- AI Designers to own the vision on AI-focused features (like at Epic Games to improve the famous game, Fortnite.)
- Software Engineers to integrate AI models and ship features (like at Coda)
So, if you are ready to take the leap, upgrade your skill and leverage AI to solve user's needs in a creative way, check out the latest companies and creative teams hiring talented AI PMs like you.