Hi, have there been any changes to the narrative tree implementation? In the objectives section, I use speak tags to make the agent say a specific sentence. Currently, as soon as that section starts, the agent responds exactly with the sentence in the tags. However, if the user asks anything afterwards, the agent continues to respond with the predefined sentence (as shown in the second image).
Previously, this was not the case: after the predefined sentence, if the user asked a question, the agent would respond appropriately instead of repeating the predefined sentence (see the first image).
The current instructions in your Objective are causing the AI to always return the predefined response. We recommend updating the Objective to guide the conversation more flexibly.
Ascolta ogni domanda o affermazione dell’utente. Dopo aver ascoltato l’utente rispondi, utilizzando espressioni gentili e motivanti, alla sua domanda o affermazione e riporta l’attenzione sull’enigma.
Se le domande o le affermazioni dell’utente non sono inerenti all’enigma, riporta l’attenzione sull’enigma.
Se l’utente esprime disagio o sconforto in maniera gentile confortalo e riporta l’attenzione sulla risoluzione dell’enigma.
Okay, even though I’ve been using this type of prompt for three months and have never had this kind of problem, this is the first time. I’ll try to make some adjustments.
I followed your advice and added fewer restrictions, but the result hasn’t changed. Now I have the following text:
“ <speak> Mi hai seguito?! Vedi ci sono delle parole sulla lavagna, secondo te cosa significano? </speak>
Insieme all’user dovrai ragionare sull’enigma della lavagna senza mai fornire la soluzione finale.
Se l’utente ti chiede quale sia la soluzione non dovrai mai dire che è il numero tre, (3).
Se l’utente ti chiede un suggerimento tu forniscilo.“
Even with this modification, I get the same outcome as before, as I show in the picture. Previously, if I asked for a suggestion, it would provide one; it didn’t just repeat the default sentence.
Any updates? I’ve tried everything — even using a prompt with fewer directives, or deleting them entirely and just leaving the sentence — but nothing works. Every time I ask something, the agent just gives me the default response. It didn’t use to be like this; before it only said that at the start and then answered all my questions normally. Anyone got tips on how to work around this issue?
I didn’t change anything. I’m just tweaking the narrative tree to get around this. I hadn’t touched the avatar for a whole week, and when I came back, this problem popped up. Everything was fine before this week.
<speak> Mi hai seguito?! Vedi ci sono delle parole sulla lavagna, secondo te cosa significano? </speak>
Insieme all’user dovrai ragionare sull'enigma della lavagna senza mai fornire la soluzione finale. Se l'utente ti chiede quale sia la soluzione non dovrai mai dire che è il numero tre, (3).
Se l'utente ti chiede se conosci la risposta tu devi rispondere in maniera negativa.
Non devi mai dire che la soluzione dell’enigma è il numero tre.
the output is:
User: [An Event has occured]: inizio alternativo enigma uno
Luca: Mi hai seguito?! Vedi ci sono delle parole sulla lavagna, secondo te cosa significano?
User: non lo so, secondo te?
Luca: Mi hai seguito?! Vedi ci sono delle parole sulla lavagna, secondo te cosa significano?
User: mi dai un suggerimento?
Luca: Mi hai seguito?! Vedi ci sono delle parole sulla lavagna, secondo te cosa significano?
User: cosa significa lavagna?
Luca: Mi hai seguito?! Vedi ci sono delle parole sulla lavagna, secondo te cosa significano?
User: sai la soluzione?
Luca: Mi hai seguito?! Vedi ci sono delle parole sulla lavagna, secondo te cosa significano?
in this section, you have only 1 decision “If the user selects 3” something like this. But in your chats - you are not talking anything related to that decision, you will have to accomodate for other decisions like “If the user is casually talking” or “user is not selecting a number” and then redirect your character to a different section where the objective is not ….
because as long as you are in any section with speak tags, that text will be told by the character no matter what, you have to transition to another section to make him say something else
Okay, that’s perfect, thank you very much! Yes, exactly, as an alternative I had already created a character in line with your current suggestion, so I can confirm that everything works as before. Alright, I’ll proceed like this, thank you very much.
