Category: Chat

I’m curious how people feel about this idea in general.

Imagine an AI fiction app where both people keep their own completely private account.

Your stories, chats and characters stay yours.
Your partner’s stay theirs.

But there’s also a third space you can create together.

If there’s something from your private side you actually want to share, you choose to bring that specific thing into the shared space. From there, you can read it or continue it together.

So the model is basically:

Your world. Their world. A shared world.

No shared login.
No automatic sharing.
No access to each other’s private activity.

Would that make an app like this more interesting to use with a partner?

Or would you rather keep this kind of thing completely private?

Hey guys! Has anyone here tried meta AI for a business page’s messenger chatbot? Is it customizable like what you can do with n8n? If anyone can provide pros and cons, that would be great!

Hey guys! Has anyone here tried meta AI for a business page’s messenger chatbot? Is it customizable like what you can do with n8n? If anyone can provide pros and cons, that would be great!

This is Kevin. He’s your professional rival, you’re competing for the same career opportunity, and thanks to one stupid decision everyone now thinks you’re dating.

His rules are:

No jealousy.

No getting attached.

Whatever happens while pretending doesn’t count.

I tried to make him genuinely competitive and independent rather than instantly obsessed with the user. I’d really love opinions on whether the premise/profile actually makes you want to click him or whether I’m missing something.

Character link: Kevin

Hi, i want to create a rag based chatbot without llm, it should be faq based but my source of info is very limited like 300 questions so i dont think so it will be requiring LLM, now i have used tf-idf and fuzzy score and i want it to be bilingual, it might sound like i can give a prompt in gpt but i want to explore options, it should be production ready, and doesnot require llm currently, maybe in future i can implement, so what approach should i use? It does have flow based queries for faqs like steps to solve different errors etc. I have implemented flow based approach but i am not able to get the gibberish queries out and if it cant answer texts we can just raise a support ticket, and how to test it too, also i thought of using embeddings but idk how to

My partner and I run a small shop and we keep missing calls during busy hours. Someone mentioned AI receptionists but I have no clue where to even start. Anyone actually using one of these?

I was role playing like usual, but when it got to a nsfw scene it backed down saying it cannot continue. Which is weird cus it always allowed said scenes.

For context it is a dark roleplay but like it’s been allowed. Just now it had that happened which is strange

Ok guys, we all have one now… maybe 3… maybe 5…

Which one are you most proud of? Which one do you think will make the cut…?

I was looking for a ai character chat platform that is unfiltered, I was wanting to see what people think the best ones are, I’ve used silly tavern which is a lot of setup, I don’t have a strong PC to run local, and I’ve used emochi and zeta which are mostly unfiltered

I’m building a personal AI assistant as a learning/project thing. It has persistent memory, tasks, reminders, goals, events, and a classifier that turns normal messages into structured JSON before anything gets written to the database.

I recently switched the classifier to the Gemini API (currently Gemini 3.6 Flash). It actually performs pretty well so far — for example, it correctly extracted a goal, reminder, and event from one message and handled relative dates correctly.

The problem is reliability. During testing I’ve already hit multiple 503 UNAVAILABLE / high demand errors. I added exponential retry/backoff (2s → 4s → 8s), and on one test it failed twice before succeeding on the third attempt.

I’m still early enough in development that changing providers wouldn’t be a disaster.

For people who have built similar assistants/agents: would you stay with Gemini and design around temporary outages, or switch to another provider before the project gets more dependent on it?

I care more about structured-output reliability, availability, cost/free-tier limits, privacy, and API stability than having the absolute smartest model.

Also curious what you’d use instead and why — OpenAI, Anthropic, local models, or a multi-provider/fallback setup?