Every day I notice AI tools integrating more seamlessly into my personal and professional life, and I have been wondering if there are better ways that I could leverage new AI tools in my work as a product manager. To answer this question, naturally, I asked ChatGPT.
My prompt was…
I’m a product manager at a financial and technology company, and I specifically work a lot with data and data infrastructure, as well as ML. How can I best take advantage of new AI tools in the world to improve my efficiency and effectiveness at work?
ChatGPT emphasized that my role as a product manager stands to benefit more than a lot of other roles from the use of AI, as it can act as a thinking partner, translator, and force multiplier (which, when I looked it up, means multiplies the effectiveness of a work entity). The following are specific ways that it suggests a PM can take advantage of AI tools to become a better product manager, along with my thoughts on each:
Product Discovery Synthesis
Input raw interview notes, Slacks, etc. and ask an AI assistant to summarize pain points, extract potential opportunities, and surface assumptions.
Note: Be aware of data privacy and sharing concerns here.
PRD Drafting
Give an AI assistant the problem statement, constraints, and target metrics, then ask it to draft a PRD with goals, edge cases, and risks.
Note: This one would require a pretty extensive prompt to give the AI assistant enough context to make a useful PRD. Maybe focus on just a few sections of the PRD instead.
Brainstorm & Explain ML Solutions
An AI assistant can help a PM translate ambiguous or complex business requirements into concrete model requirements or specs. It can also help translate an ML solution into stakeholder-friendly language.
Note: I would also love to use an AI assistant to brainstorm potential input features to a model given the type of data available at the company.
Data & Anomaly Analysis
Paste a data table or schema and ask it questions about the data, like what business concepts it represents or what type of queries you should run on it. Alternatively, describe a KPI and recent anomaly and ask it for possible causes and data freshness checks.
Note: Has anyone ever tried to submit something besides plain text into an AI assistant? Need to try this.
Craft SQL Queries
Input a rough query or natural language request into an AI assistant and ask it to turn it into a proper SQL query.
Note: Alternatively, you could ask AI to validate your SQL query, or help de-bug it.
Design Data Infra
Use AI as a thinker and designer for data infrastructure, including data pipelines, feature stores, or ML pipelines. Specifically, ask it to outline tradeoffs of different solutions.
Note: This seems like it might be more useful in assisting engineers rather than PMs, though if I was going to use AI like this, I would also use it to try to get a rough LOE.
Comms Drafter
Ask AI to draft the initial version of a short comm, like an executive summary, product FAQ, or launch announcement. If you can, ask it to tailor the blurb to a specific audience, or write it in a specific tone.
Note: I’ve already started using this and I’m a huge fan.
Limitations & Warnings
AI assistants are good for some things, and really bad for other things. Generally, AI is good for expanding options or a solution space, coming up with additional risks or considerations, and automating manual tasks.
AI assistants should not be trusted to make final decisions or conclusive judgment calls. You should be wary of numbers or metrics, and you should be careful about trusting any advice related to compliance or regulations. Don’t ask AI to make prioritization decisions, as it is likely missing part of the context important to the decision. Also, aim to use AI to develop initial drafts of pieces of writing, but always review and edit before using.
I would love to hear any other warnings you have about utilizing AI as a tool for product management – comment your ideas!
Your message has been sent
Thanks for reading!
BYOMSPM 
Leave a comment