Learning from Product Firing: CreatorMind
When we unsubscribe for from a product, we often call it firing. Today I officially fired two products: (1) the advanced version of ChatGPT and (2) CreatorMind.co. I’ll address the ChatGPT in another article for this article I want to focus on creator mind.
Big picture
The reason I’m firing this product shows or illustrates how difficult it is for startups to compete as well as how far we still have to go with AI tools. I needed a broader product but the technology is currently better at more specific use cases. (Think deeply vertical solutions for expensive problems.)
The User’s Goals
Let’s start with the user perspective.
I was paying $15 a month for about 25 people to ask questions over 4 months, it ended up being $3 per query!
Sure it was fun to see the questions people were asking. However, the technology couldn’t learn enough from my articles to answer their questions as quickly or specifically as they wanted. Sometimes that was because it didn’t have enough information, so they were getting mediocre answers and getting frustrated.
But often it was problematic because the questions weren’t specific enough, my users weren’t having deep conversations with the bot. It wasn’t about chatting; the users wanted an “answer machine.” Some folks wanted the answers handed to them, others were testing my skills to see if they wanted private coaching from me.
Additionally, the questions they were asking were all over the place. The tool can only do one thing. It can read from my articles. However, my articles are not written to make it easy for the machine. I bet if I filled out a form to answer the common questions, it could do it, but that’s a different type of chatbot.
Which gets us to the core of the problem. At this stage, and we were even at this stage with chatbots long before AI came along, you need hyper-specific solutions for different types of prompts. And the solution that could search my general articles, and give an answer, is different than a chatbot that can answer customer queries.
Use Cases
How were my chatbots being used?
Customer Support or Customer Service Queries
i.e. I have a question about something I don’t think is working as it should on your website, so I’m gonna ask the bot thinking it is a real person.
E-mail Replacement
Instead of sending an email, trying to ask the chatbot as if they really think they’re talking to me, again.
Complex Questions
Asking questions to complex questions that have multiple possible answers or answers that need to be personalized
The Questions
The top five types of questions were:
Frameworks
Product Sense
Specific Prompts
Customer Service Requests
General Interview Prep
Using the Data
How should I use the data?
It was expensive for what I got. But it wasn’t entirely a loss. And I might come back. But that doesn’t mean I didn’t get some value from it. I will use the data to:
Writing articles directly addressing those questions as worded and see if that helps with SEO.
Improve tagging of my articles
Highlight items in my blog
Try building a reputation on LinkedIn for answering the questions asked.
Summarize
It is fun to test AI products in today’s world where a new idea comes up daily. But move forward with caution; there is still a long way to go for real solutions that increase sales and/or productivity. We are getting past the hype to the “show me the money” stage.
