Hi everyone,
I’m using Base44 connected to OpenAI’s to run a controlled experiment with two personas:
“Human manager” and “AI Manager”
Both are powered by ChatGPT via Base44, with prompts carefully designed to guide their tone and behavior.
When I test directly in ChatGPT, the personas behave exactly as intended.
But in Base44, the personas are far from perfection with really poor quality, as if they have no idea what’s going on, detached from my
I’m looking for help getting Base44 to replicate ChatGPT’s behavior more reliably, or suggestions from anyone who has faced and fixed similar issues.