I am too controlling of most things to hand them off to AI. For instance, writing this article or writing code. I want my writing to be in my voice and I am still learning to code, so no taking shortcuts there. But, I have found AI to be helpful and practical in automating lots of things. I wanted to share some of these in case they spark ideas for others on how to use AI in practical ways each day. Let's get into some examples:

I help to run a local product management group. We use Slack as the hub for our community. Our members requested that we have a weekly topic to start discussions in the group. I took it upon myself to post those conversation starters. But I ran into two problems. It turned out that I wasn't that good at coming up with topics. I was even worse at remembering to post them. So I turned the job over to our Slackbot. I gave it access to the GPT API and instructions on the types of questions to ask. Now each Monday morning it runs a job to generate and post a discussion topic. It never forgets to post and is much more creative with questions than I could ever be.

One of the best ways that PMs can create change in their products and careers is to spot trends and innovations. To be able to see the spark of something new and find a way to seize upon it. There is a ton of product management writing out there (I would guess I am subscribed to 1000+ newsletters at this point). But, it's frustrating that most of it is focused on what has happened in the past or on general strategy. I wanted something that could tell me what was going to happen and what I should do about it. I couldn't find anything. So I built it myself. I created a daily, AI-generated newsletter. It goes out and reads all the latest news. Based on the instructions I gave it, it decides what news is relevant for PMs. Then it summarizes those items and sends them straight to my inbox. It was way more work to build than I expected but in the end, it has saved me 100s of hours of reading. More importantly, spotting even one relevant innovation is a huge ROI. For example, a while back it surfaced this article on weather forecasting from DeepMind. Weather forecasting is compute-intensive, expensive, and challenging. Most companies who rely on weather outsource the work and buy data from a handful of vendors. But DeepMind has found a way to do it cheaply and easily. Now companies who need weather forecasts can build them in-house. It's a chance to save themselves money and opens up innovation in how they serve their customers.

To OpenAI's credit, they have some of the easiest APIs to work with. If you know even a little bit of Python you can start to work with their AI models. When they released a new text-to-speech model, I was excited to give it a try. Particularly I had been looking for a way to create audio versions of some of my blog posts. This new model seemed like a great way to try that. I started with a Python script and within 20 minutes, I was feeding it text and getting back mp3 files. But I wanted something more scalable and accessible. So I took that initial script and built it into a web app. Now in a few clicks, I (or anyone) can generate an audio version of almost any text.

I have mixed feelings about sites like G2. They are a treasure trove of data on companies, but every company seems to have a 4.8 out of 5 rating on there. The veracity of the data sourcing aside, I wanted to see what insights might be hiding in that data. But I didn't want to read through hundreds of reviews. So I wrote a Python script to do it for me (I probably need to write a separate post on the high leverage for PMs in learning to code). 

The script asks for the product and a category. It then grabs all the reviews for that selection and feeds them to an LLM with these instructions:

I am a product manager at a company called SuperHR. The data contains reviews from our customers. I am specifically interested in the reporting part of our product. Using the data provided can you give me a few things:
- Summary of what customers like about the current reporting options
- Summary of what customers dislike about the current reporting options
- A list of ideas for possible functionality we could introduce to make our reporting tools more innovative (for this one use the data provided and your understanding of business intelligence and HR tools)

Anytime I need to research a new feature or gather customer sentiment, I can run this script and get a set of curated data to work from. (This one does require access to the G2 API but you can also get around it by downloading the reviews manually).

I once worked on a partnership where we needed to integrate a product with thousands of database tables. It was my job to decide which tables we needed to pull data from. But I had no idea what was in any of the tables or how they related to each other. I spent a few weeks exploring that data trying to figure out the answer. But now I can get those same answers in a matter of minutes. Using a combination of Python, SQL, and LLMs you can have an AI agent inspect an entire database for you. It can describe the the data in each table, and give you suggestions for how to use that data. Even more helpful, it can also tell you how the tables relate to each other and how you might use the joined data. Now I can onboard a new database in a matter of minutes and become an expert in it within a few days. You can try out a sample version of this concept here. (I am super excited by this thread and am diving into it deeper. Soon I will have AI detecting PII for me, generating ERDs, and recommending ways to combine data sources. There is tons to be gained from teaming up with AI in the database domain.)

A similar problem to the above, have you ever started a new job and been handed a set of obscure SQL queries? You have been told they are important and that you should use them, but you have no clue how they work or what they are for. This has happened to me several times. Usually, I would spend a bunch of time figuring out what the queries do. But not anymore. With a quick script and API call, AI can do it for me. I can give it a complicated query and it can explain what it does in an easy-to-understand way. Not only that, it can also optimize the query and rewrite it to perform better

So far everything I have covered requires that you can write code. But what if you don't know how to code or don't want to build something new? GPTs are another way to automate a lot of the repetitive work that you do. I created a GPT that can help me in my role as a PM. I can give it a new feature or product that I am considering. Then it will: brainstorm with me, do market and customer research, compile and chart data, provide feature ideas, and finally compile everything into a document. It isn't perfect or as good as a human but it saves me tons of research time and lets me focus on the highest-value parts of work.

Automation isn't only for repetitive tasks. You can also hand over one-time time actions to AI. We recently partnered with Verdi on a PM x AI workshop. Part of the event was a pitch competition. Rather than us picking the winners, we let AI do it. We built a GPT that knew the rules of the competition and the qualifications for each prize category. We then gave it summaries of all the pitches and let it pick the winners. We could have judged the pitches ourselves, but this approach took the review time down to a few seconds and removed our bias from the decisions.

I'm trying to use AI in practical ways all over the place. Hopefully, this has given you some ideas for how you could do the same. If any of these caught your eye and you want access to the code or need help getting started please reach out. Or if you just want to talk shop, my inbox is open.