ok, im also getting that same hiiiiiii response with this now as the prompt for personality agent You are a helpful, respectful and honest assistant. Always answer as helpfully as possible.

hmm can you create a report id? i'd like to take a look at whats happening in the payload

just reloaded.... im getting repeating messages again...

i think the personality prompt is throwing this in

{ personality: 'You are a helpful, respectful and honest assistant. Always answer as helpfully as possible,', before: '[{"type":"text","text":"Gooooood afternoon! I\'ve had too many carbs for lunch, that wasn\'t a good idea! These clouds make it feel so dark and gloomy but its real hot, hot enough for a bbq 😉 It\'s 20.91°C here in London!","typing":true}, {"type":"text","text":"Anyway... let\'s start by taking some details and figuring out what you are looking for... my name is Lucy by the way, **what\'s your name?**"}]', after: [ { type: 'text', text: 'Good afternoon! I understand that having too many carbs for lunch may not have been the best idea. The weather seems a bit gloomy with dark clouds, but it\'s still hot enough for a BBQ. The temperature in London is currently 20.91°C. How can I assist you today? By the way, my name is Lucy. May I know your name?' } ] } {"scope":"agent-hook","agentId":"PersonalityAgent","hookName":"conversation_turn_before_reply"}

i've never seen this before. Can you open up a report-id, i'll look into the merge and open a ticket correspondingly

55f42c2e-d167-45b6-aab5-8cd38c6303e3

I can speak to that, not on the core engineering team

Yes, why, is there a problem?

yeah seems like im struggling to load it. let me find someone to help from engineering

That's strange, sorry i know you probably don't need this issue right now

i load these reports 5-10 times a day, first time it has failed me, all in all i'm still pretty happy with our problem report feature

Ok, let me know if there is anything i can do. does personality agent send the text to gpt and rewrite and send it back?

yeah

that must be so expensive, why dont you guys build your own llm off the back of llama 2 or another opensource llm

Yeah, we expect the cost to go down drastically and performance to up massively as well on so many models