ChatGPT: A PM's Best Friend 🤖
Five ways to embed ChatGPT into your PM workflow, plus other AI productivity tools to amplify your craft
The best way to learn more about generative AI, its potential impact on society, and how it can benefit you is to try it yourself! ChatGPT is the most popular way to interact with GPT-3, a large language model that simulates real-life conversations.
As a PM, here are five use cases you should explore:
Understanding complex topics as if you are speaking with an expert
Get help writing that SQL you’ve been meaning to brush up on
Write better PRDs
Improve your copywriting
Land a new job
Use Case 1: Understand complex topics with ease
This is my favorite use case! If you’ve ever gone on a wild goose chase through white papers (repeatedly smashing command F), Wikipedia articles, Reddit, or a series of coworkers trying to understand a tricky topic, then you’ll want to give this a shot.
Example: Understand what factors influence the amount of rewards the Solana protocol gives out to validators.
Key Insights:
While ChatGPT isn’t up to date with the latest or most accurate information, its clean UI and conversational style make it helpful for understanding complex topics - like you are discussing the topic with an expert.
For contrast, here’s the first result that Google served up when I queried What factors influence the Solana protocol APY? There’s a lot of ads, images, and irrelevant information that I need to comb through to find a clue to my answer.
With Google, it may be a bit easier to make a judgment call on the validity of the source. For example, I know the official Solana website is a trustworthy source, while Reddit is more questionable. With ChatGPT, sources are more opaque. Some tricks to improve the accuracy and verifiability of its responses are:
Encourage the model to use logic and reasoning to get to a result. For example, asking it to respond in the format Let’s think step by step. <give_reasoning> Therefore, the answer is <final_answer> has been shown to improve the accuracy of responses. This is called chain of thought prompting, and is the subject of a few papers on the subject such as this one. Check out the difference in output between the first and second prompt below.
Ask the model to cite its sources! In the response below, I was happy to see the official Solana staking documentation and white paper as sources.
Use Case 2: Write that SQL you’ve been meaning to brush up on
Example: Write a SQL query that outputs the top 5 countries by trading volume on Coinbase on Christmas Day 2022.
Key Insights:
Half of the work of a SQL query is understanding what data is stored in the tables you are working with. Once you’ve done that, it’s off to the races!
Use Case 3: Write better PRDs
When laying out the requirements for a product, it’s helpful to include an opportunity section that lays out the reasons why we are building a certain product. It’s also important to contrast the opportunity with the potential risks. Building strong mitigation plans to those potential pitfalls make a much stronger case for your product and smooth execution.
Example: Weigh the opportunity and potential risks of launching a subscription-based membership program at Coinbase that would include zero trading fees.
Key Insights:
I’m often a bit biased towards spending more time on the opportunity section and the potential gain that our launch will result in. However, the risks are equally as important. Use ChatGPT to amplify your thinking around the risks or highlight interesting areas you can dive into more deeply!
Use Case 4: Improve Your Copywriting
Example: Make a FAQ section on the Coinbase website more understandable to a beginner user.
Key Insights:
Running copy through ChatGPT and highlighting the goals of the text can be quite enlightening. Although the original FAQ in this example is written well, I’d argue the output given is more accessible and could be a better choice depending on the target audience.
The output of ChatGPT is not deterministic – in order words, you can run the prompt multiple times and receive different responses. This may help with writers’ block!
Use Case 5: Land a New Job
Some ideas on how to use a Chatbot in your search for a new job:
Write a more concise, powerful cover letter or letter to the hiring manager
Practice product management interviews
Understand product frameworks
Example below: Practice a product execution interview question.
Key Insights:
Leveraging a LLM can be a great way to get the ball rolling on emails and messages that you may not be used to writing – just remember that ChatGPT doesn’t know much about who you are, so ensure all of the information is accurate before using it.
It may be helpful to provide a few examples of questions you know have been asked at target companies to enable ChatGPT to ask more realistic questions
Other AI Tools We Are Excited About
What it is: AI-powered storytelling. Type in a prompt and Tome generates a multi-slide presentation, complete with AI-generated images in the style of your choice.
How to use it: As a kickstarter for your pitch or presentation. I tried creating a presentation outlining a trip to Singapore and Vietnam and this is what I got within a few seconds. If you want some extra credits to play around with the tool, here’s my referral link.
What it is: your very own AI-powered AP English teacher in an easily accessible web app ;)
How to use it: If you have an important slack message or email that you’d like a second set of eyes on, paste it into the Hemingway website. It highlights hard to read sentences, weakening phrases, use of passive voice, and more.
What it is: an iOS app developed by Quora that allows you to easily interface with a number of chatbots, each based on different LLMs.
How to use it: Try out the various LLMs such as Claude, Sage, and Dragonfly. Claude is the LLM developed by Anthropic, which recently received $300M in funding from Google to develop more reliable, interpretable, and safe AI systems.
This is super helpful, I'll be using these tips. And I like the curated section of other cool AI tools.