Facts feature | User facing understanding

[josancamon19]

Is your feature request related to a problem? Please describe. We already have a generated list of facts of the user based on their conversations, in the facts subcollection.

We need to a few things with it

• [ ] ~For some users with so many memories, they'll get 100s of facts, maybe reduce them with llm or something to condense them~
• [ ] ~Facts should be cached in backend.~
• [x] Prompts in backend should access user_name + facts directly from function prompt, not as a dependency.

---

Moved to #395

• [x] ~From settings it should be available for the user to set a name -> This should be set in firebase auth, and trigger a refresh of facts "The User" or previous user name replacement, so that facts are "migrated" and use new user name.~
• [x] ~Name probably should be asked in onboarding too (Some users hide it with apple) or maybe have a business account with not their name.~

---

Users should be able to review facts

• [x] There should be an UI in the app that displays the list of facts extracted from user conversations.
• [x] The users should be able to review them, similar to bee computer tinder ish, but the value should be clarified to the user, this will help AI be more personalized.
• [x] Backend should use reviewed as True, and reviewed as False, should be set as "Kim likes x" Kim doesn't do x"

---

Smarter Facts

• [ ] We need to find a way to find basic characteristics about the users, like, where is it, location? Sex (ask it?). Work, age, etc...
• [ ] Try more directly related retrieval of facts, like basic + context related topics using memory_category field.

[josancamon19]

Facts Part 2

• Conversation on chat should allow the llm to verify/discard/add new facts about you (tool function).