What separates a successful AI product from a failure?
Adam Gusky
9 replies
Replies
Philippe Wellens@pwellens
Having worked for C3 AI in Products and AI for 6 years, I can tell you that a lot has to do with trust and understanding of the AI insights and outcomes. This is particularly true for AI software targeting businesses. It also comes with change management. The key is to address this through the application functionalities.
Share
I think, like any other product: 1. the buyer needs it 2. it's user-friendly and well done 3. you know how and whom to sell it.
leonardo da vinci
@adam_gusky Accuracy vs Inaccuracy. An AI product is a turnoff when it doesn't produce expected results.
Vidiofy
@adam_gusky @aditya_chavan3 Love this Aditya. Especially when you feel that the AI tool "just didn't get it". Though I would say that due to the hype around AI, expectations can be a little too high expecting a perfect output every time.
@aditya_chavan3 yeah, In our experience, if the client see direct model output, you need to have accuracy over 90% for them to think that it semi works
leonardo da vinci
@louisa_lu1 in B2B domain the clients are but more patient and will share a feedback. I have been launching B2C AI products and the users are absolutely ruthless 😅.
Actually finding product market fit. At the end of the day, users dont really care whether a product has AI or not. It should server to solve some problems (or great vitamin). A lot of products are hard to tell on the AI part, for example, midjourney is raking in hundreds of millions, the product is soooo cool but I don't know a consistent use case (because i dont have the data). Another case would be character.ai, which I feel we are sort of competing again in some way. User stats for character.ai might be stronger than early facebook stats, but monetization and usecase is not very clear.. although I am super bullish on both companies!
leonardo da vinci
@louisa_lu1 do checkout messengerx.io they competing with Character.ai. thier ui/ux isn't that great but the product seems to be solid.
@aditya_chavan3 ah interesting, thanks for letting me know