My lauch report for https://t.co/iMewNdfgvW. I talked about:
- How we have a decent conversion rate but $0 revenue.
- How we ranked on the first page of Google for certain long tail keywords
Read here: https://t.co/0KNxpNUQjF
#buildinginpublic#indiehackers
@MrNick_Buzz Have 2 clients for https://t.co/cKgodWXZg8. Currently both are pre series A startups.
Both have a lot of requests and negotiate price quite hard.
Want to hear your advice on getting more startup clients and how to slowly transition to other less resource intensive clients.
No easy way, I've been trying for the past 2 months.
At first: a few promo posts => those quickly got removed by mods
Then:
1. Find 5-10 top contributors
2. Interact with their content
3. Once they interact back, DM them for a quick intro & ask them to trial out my tool
4. Getting their feedback & if they are happy, ask them to help me post a review on the group
@kimelnick Yeap this approach fits when your business has a lot of user generated data that is safe to publish.
Slack wouldn’t make much sense because leaking private data is a disaster for them
It's a painful story but as with any startup story, there's brilliant piece in it:
"SuperPrix didn’t allow us to put our logo on their checkout ( which we did later on other stores ) - but they didn’t say anything about the code.
So we spammed our page source and console with marketing messages. And it people found it! "
GPT is not good at keeping to a length limit (i.e 1000 characters).
However it doesn't mean you can't force it.
Simple #promptengineering trick (though might be costly if your limit is huge):
Building on top of Amazon Product API and realizing how ridiculous it is for a dev.
To sign up: need a partner account with min 3 sales.
To maintain access: make 1 referring sale every 30 days.
This is why products like rainforestapi exists.
So, how much do they make?
Here's a guess:
Traffic: 42.6k/mth on similarweb
Base plan: $59/mth
Rev: from 1-10% conversion rate: $25k -$250k MRR
Now which other big businesses have ridiculous API requirements?
If you still think that prompt engineering has no moat just check out this story:
LexisNexis acquires CaseLex for $750M in exchange for 8 OpenAI prompts that give good responses to law questions.
@ramsri_goutham Can’t agree with 1 more.
Recently had a high hallucination rate in my app due to this automatic upgrade behavior.
It seems like the push to higher token limit creates a trade off in generation quality
TIL: Some biggest Onlyfans creators outsource chatting (their main money making machine - not selling pics/vids) to Philippines.
So it's lonely dudes chatting with dudes in third world countries pretending to be the star? 🤯
@HyunGunJung Can't agree more.
Killed my first 2 products because although i was solving my own problems, the market perception pointed to vitamin issues, not pain killer ones.
Hey Dan! Tbh my final 2 options come down between FusionAuth and Clerk.
In a different setting (e.g when I had a team and at least 1 devops), it would have been FusionAuth over Clerk.
However being a solo founder myself, the most important thing is speed of integration and subsequently feature development on top of the auth platform.
Encountered a strange case of hallucination due to very long prompt on gpt (1k tokens)
Turn out the solution is to split that massive single call into smaller calls and just concurrently resolve them before patching the content together.
The whole generation got faster as well due to async.
#promptengineering