Practical AI for PMs | 04.04.24
It’s been a tumultuous week in AI – but that’s to be expected these days. Hi there.
Ed here.
As the NYT podcast Hard Fork put it on last week’s episode, The Big Guys on the block (Google, Microsoft, Meta, etc.) are buying up all the middle-class AI companies.
When I heard that statement, it deflated me a little. OK, it deflated me a lot.
But then I remembered that most of the best use cases for project managers don’t involve developing new or novel AI models. They involve putting innovative wrappers around these models to solve real-world problems for busy people like you and me.
Google didn’t build its own internet; it was built on top of the one that existed. The same is true of Facebook and all the rest.
So just because the top companies in AI are hoovering up the top AI talent (and they are), that doesn’t mean smart, motivated people aren’t still out there leveraging AI to build cool new tools to solve real-life challenges for project managers.
New AI startups are popping up out of garages (more likely post-COVID home offices) to solve small but interesting project management problems every day.
I hope this trend continues despite the consolidation at the top. Then, maybe we won’t be forced to accept a few boring, gimmicky “AI” features developed over a long weekend by the major PM SaaS companies so they can launch a marketing campaign about being relevant in the AI space by Monday morning.
Here’s 🤞hoping!
Reading I Found Interesting: Demand Increases for Project Professionals with AI Skills, Yet PMI Research Finds Only 18% Have Practical Experience
Community-Led AI and Project Management Report
Artificial Intelligence Specialists Share Thoughts on Recent Community-Led AI Survey
This Week’s AI Use Case: Custom Knowledge Base Agent For the past few weeks, I’ve used a tool like this to remove friction from my daily workflow. I’d love to share the live version with you, but see Main Hurdle #1 below for why I can’t.
Pain Point:
Requiring a specific technical answer quickly, without bothering a human SME.
Solution:
Give an AI API access to a predetermined custom Knowledge Base and have it return natural language answers to natural language questions from the user in a simple chat format.
Tools:
Voiceflow.com (Workflow Platform)
Open AI API (AI Agent the Workflow Calls on)
Prompt Tips:
1. Be explicit and specific on the subject matter the AI is an expert in.
2. Make sure to prompt the AI to ask clarifying questions if necessary.
3. Give the AI guardrails. Require that it only use the provided Knowledge Base and none of its previous knowledge. Require that it not make up facts.
Main Hurdles:
1. Getting authorization to use proprietary information in an AI query.
2. Time spent verifying results based on the accompanying citations the AI includes in its response.
If you’re interested in building your own Knowledge Base ChatBot, just reply to this email and let me know. I’d gladly walk you through how I did it so you can apply it to your specific use case.
OK. That’s all. Talk to you next week.
-Ed
If you found this newsletter valuable, please consider sharing it with a friend or colleague. Here’s the link where they can 👉 subscribe .
AI Disclosures: The content of this email was written entirely by me, Ed, a human. This email contains NO AI-generated content.
Powered by beehiiv