RPA Challenge done with ChatGPT

Goodbye selectors, goodbye. You will be missed, at least by me :(

What you can see on the video is RPA Challe nge solution that runs without hardcoded selecotrs. Selectors are being generated by GPT3.5 based on the HTML content and it's 100% accurate. Right now it's a little bit slow, at least for RPA Challenge, but combined with fine-tuned Llama 2 7B models it will run much faster. Such capability will be implemented in every RPA tool soon, and then selectors will be gone. This is of course an experiment, but I have played with GPT generated selectors quite a bit and possibilities are endless. Automations that self-repair and that works dispite changes to undelying application. GPT integration with CI/CD pipelines, so some CRs are implemented in solutions automatically and last, but not least, text-to-automation RPA solutions. And I'm definatelly not alone in such thinking. During Ai4 confernece UiPath presented theirs approach to text-to-automation solution called "Project Wingman". I'm definatelly looking forward to test it. But, meanwhile, please let me know if you would like to see some of these explained in more detials.

[arslanimtiaz]

Its fantastic, can you upload full detail video for it.