35 user interviews later: I stop building Noble Answers, the Stack Overflow for No-Code will not work. Here are my learnings. Thread 𧡠πππ
#buildinpublic #failinpublic
#buildinpublic #failinpublic
(1) Core idea worked: Stack Overflow for No-Code. Video snippets vs. code-snippets (400+ sign ups organically)
(2) Core product idea video snippet answers vs. forum posts (save time, learn faster) worked (user interviews)
(2) Core product idea video snippet answers vs. forum posts (save time, learn faster) worked (user interviews)
(3) Biggest challenge: Technology. I used GPT-3 from OpenAI and got it working. But it takes 14 minutes (roughly 40 queries to get to the right answer). When I started the product I expected it would be 10 seconds or self-service.
What does this mean for the business:
What does this mean for the business:
14 minutes = 4 answer per hour per expert. Expert rate at scale $200. Therefore 1 answer is $50.
Stack Overflow has 21 million questions and 31 million answers.
It would cost $1.5B to get to Stack Overflow scale.
Stack Overflow has 21 million questions and 31 million answers.
It would cost $1.5B to get to Stack Overflow scale.
Let's say total funding amount for Noble would be $200M and let's say $20M into answers over time (10%). Current cost: $1.5B, goal: $20M. Therefore I'd need 75x cost improvement.
Instead of 4 answer per hour, 1 expert could do 300 answers an hour.
Instead of 4 answer per hour, 1 expert could do 300 answers an hour.
To get the quality necessary to provide value, the answer accuracy has to be world-class. It's unrealistic that I get to this sort of performance improvements with off-the-shelf AI from OpenAI.
(4) I'm super proud how much I learned with Bubble/no-code. I did not need to raise money (people offered me to raise $2M for $8M pre) on the idea. But I'm super happy that I could validate the product and have the freedom to pursue another idea.
(5) The opportunity cost of pursuing an idea that is not working is just too high as an entrepreneur. I'm super proud that we validated the business in just a few months. And I'm even more happy that I did not loose money of an investor. This is clearly the future.
(6) No-Code will allow us to have 10x more shots on goal β½οΈπ₯
. Traditionally it took me 18 months (540 days) to validate a billion dollar idea. I think I can get it down to 1.5 months (54 days).
(7) Thank you to all the supporters of Noble and I'm super happy to answer questions and hope that my public closure of Noble will show you that not everything is always perfect, but No-Code allows us to keep going for this overnight success! Let's go for it.
(8) PS: I joined a revolutionary startup in the πclimate tech space as a co-founder (responsible for the no-code product). More to be released soon.
β’ β’ β’
Missing some Tweet in this thread? You can try to
force a refresh
γ
