I'm beyond grateful to have access to invaluable information! Thank you! I work in a small startup where we help students find and apply for scholarships. One of the challenges we needed to overcome was matching students with the right few scholarships instead of just giving an endless list of matches. We're lucky to have a ton of data. We soon discovered that an AI-powered recommender would be the best solution for us. But before we went into this we actually built a pretty simple system based on specific rules and we gave weights to particular criteria. Of course, the recommendations were not great, but we validated that there's a desirability for this solution. Only then we partnered with an amazing external company that specializes in building AI engines to build our recommender since we didn't have the capability to do it in-house. It was a super fun experience. Now the next challenge for us to figure out is how do we keep evolving the recommender and improving it.
As someone who's new to PM and is helping a start up figuring out product centered around AI, this is great information! Would love to see more content around AI & PM
I got first prize on an AI Hackathon and we built a pretty solid AI product in a week. I think the availability of AI models and how each model excels in different areas will be crucial to building AI products. For the MVP you can use a commercially available model to get the financing for research. I'm not sure if we are at this stage yet, but I think we'll eventually have a pay to fork model on commercially available AI models as this would mitigate the risk of creating a tool and have it shut down due to other teams development.
It's not just about plugging ChatGPT API here or there, or employing a scientist. The software developers needs to get used to the intricacies of the AI models, data processing, data security, systems architecture. These challenges come not only at a financial cost, but can add many months of testing and technical debt pile up.
What insight! Loved listening to this in the morning. Question. You mentioned that there would be recordings of your course? Are these available only to the cohort to your course? The course seems like a fantastic opportunity. Thanks!
Thanks Lenny. This was great. Couldn’t find the link to her course tho. Excited to be doing the coursera MI course for PMs right now. Seems like things could be on good track 😅
Thank you for having me, Lenny! It was a great conversation (and honor).
5/10 great Qs, answers can be more straightforward & fruitful
I'm beyond grateful to have access to invaluable information! Thank you! I work in a small startup where we help students find and apply for scholarships. One of the challenges we needed to overcome was matching students with the right few scholarships instead of just giving an endless list of matches. We're lucky to have a ton of data. We soon discovered that an AI-powered recommender would be the best solution for us. But before we went into this we actually built a pretty simple system based on specific rules and we gave weights to particular criteria. Of course, the recommendations were not great, but we validated that there's a desirability for this solution. Only then we partnered with an amazing external company that specializes in building AI engines to build our recommender since we didn't have the capability to do it in-house. It was a super fun experience. Now the next challenge for us to figure out is how do we keep evolving the recommender and improving it.
I really liked the model metaphor - 3 year old brain & train process! Spot on!
Thank you for this Lenny. Marily is one of the best PMs out there! Helped me a lot when I was in the job-hunting process!!
As someone who's new to PM and is helping a start up figuring out product centered around AI, this is great information!
Would love to see more content around AI & PM
I got first prize on an AI Hackathon and we built a pretty solid AI product in a week. I think the availability of AI models and how each model excels in different areas will be crucial to building AI products.
For the MVP you can use a commercially available model to get the financing for research.
I'm not sure if we are at this stage yet, but I think we'll eventually have a pay to fork model on commercially available AI models as this would mitigate the risk of creating a tool and have it shut down due to other teams development.
Could you share some links of models, products and data that we can use for create a testing product? Thank you
Love how Marily explains everything with context. Awesome and v.v productive podcast! Kudos Lenny :)
Thanks for this content , I'm working my way up to find a P.M role and also into AI related.
It's not just about plugging ChatGPT API here or there, or employing a scientist. The software developers needs to get used to the intricacies of the AI models, data processing, data security, systems architecture. These challenges come not only at a financial cost, but can add many months of testing and technical debt pile up.
What insight! Loved listening to this in the morning.
Question. You mentioned that there would be recordings of your course? Are these available only to the cohort to your course? The course seems like a fantastic opportunity. Thanks!
Are there any other tools/platforms that PMs can leverage? Something more user-friendly than autoML?
Quite informational and inspiring. Thank you for sharing the referenced links.
Thanks Lenny. This was great. Couldn’t find the link to her course tho.
Excited to be doing the coursera MI course for PMs right now. Seems like things could be on good track 😅
Here you go: maven.com/marily-nika/technical-product-management
Thank you
Amazing!
What was the name of the newsletter apart from Tldr? The transcript is not correct I guess for this part of the conversation.
All the links are in the shownotes, I think you mean this one: • The Download newsletter: www.technologyreview.com/topic/download-newsletter/
@@LennysPodcast Thanks a lot. Great Conversation with Marily.
🖖🏻 Like that content 💯
Unbelievable how quickly mental models become outdated in this space. Warp speed.
the music is so strong, I don't ear his voice
During which part, the intro?
Not at all insightful
You aren’t insightful.
Didnt understand anything. Your voice kind of drops which makes it very hard to understand.