My day so far?
Increasing deployment % of rollout in hourly increases.
Watching our systems, increasing quota here or there…
EXCITED HOW WELL OUR SERVER SOFTWARE SCALES!
Scary numbers everywhere, I have to say.
Such a large rollout to meanwhile ca. 770,000 users is totally new and quite challenging for a small team like us.
It’s an experience by itself and while we’ve seen a lot already, we’ve not been where we are now.
I have never done that, and I also don’t know or have access to someone who has bootstrapped such a large and compex system in such a short time.
FYI, we also have 2 new server generations to benchmark ready for the production clusters… but no TIME to do it.
…and it’s also not yet needed. The TEST cluster hardware can take us to 1.5 probably 2 millions users.
…yeah you read that right - we’re still operating on TEST hardware… granted, great hardware, but still…
SO happy I made disruptive decisions on new technology almost two years for myself and my team.
Only the ChatGPT outages are an annoyance for us. Naturally with AIPRM being an add-on to ChatGPT we have a LOT of people complaining to us about the OpenAI outages.
Now we added a new dimension of confusion.
“Correlation is not Causation.”
Not just a problem in the SEO industry.
Now with a much wider audience I can say, it’s a common problem and misconception.
“I bought AIPRM and now ChatGPT doesn’t work.”
So still some users PINKY SWEAR that their purchase of the AIPRM premium plan was responsible for the ChatGPT outage.
Also even others come to us complaining about the second rebill of OpenAI ChatGPT - not having grasped that it’s a subscription they signed up with OpenAI, nor that AIPRM, Corp. is a different company than OpenAI Inc.
Luckily we have a lot of standard content for these support cases.
We’re just getting started!