Quick question what would be a “better” structure for our use case: We have a bunch of exported chat history logs from Slack Channels which contains a lot of know-how, mentioned problems and solutions, etc. Obviously those chats contain a lot of useless “fluff” that would unecomical to just dump in…

I would recommend the following approach: Process the 10 files using a “Creative” persona using a large context / large output LLM like Sonnet 4 thinking. The goal of this processing would be to “tidy” up the information and prepare it for a RAG Then using our built in upload, upload the 10 proces…

Thx, that helps! [画像] sam: 10k tokens is workable Just to clarify, are all the uploaded files injected into the system prompt? Or are they processed through the configured ai_embeddings_model first and then injected? I am a bit confused about your recommended 10k token limit recommendation …

The files in Discourse AI Persona, upload support are only limited by your upload size, they can be huge, they are processed via embedding, we inject chunks into the prompt per configuration. What I was talking about was trying to coerce all the information into a single system prompt here: [imag…

Ah, that clears is it up, thanks! So basically, my next steps should be to run a few tests with different embeddings models and see what token size I end up injected into the system prompt, right? And when creating those txt files with chunks, I should make sure that they stay within a reasonable …

[画像] jrgong: So basically, my next steps should be to run a few tests with different embeddings models and see what token size I end up injected into the system prompt, right? The embedding model controls quality, not quantity you can roll up all your data into a single file, we will chunk i…

Thx sam, I really appreciate it :heart: If you have any further helpful resources, feel free to share them here. Once I make progress, I will try and publish my experience here on meta. :slight_smile:

@sam how do you suggest adding version or model numbers to metadata separators? Your original example : [[metadata about cats]] a long story about cats [[metadata about dogs]] a long story about dogs Now if we want to enrich them with version numbers or specific model numbers, do I just use the sa…

[画像] jrgong: do I just use the same format or structure that humans would use when typing them? Yes, this is the way to go!

チャット履歴に頼るペルソナのエンジニアリング

サポート

sam (Sam Saffron) 2025 年 8 月 11 日午後 10:59 9

はい、それが最善の方法です！

トピック		返信	表示
Discourse AI Persona, upload support Announcements ai-bot , ai	21	1693	2025 年 9 月 11 日
AI bot - Chat and PM integration Site Management how-to , ai-bot , chat , ai	8	1158	2024 年 10 月 24 日
Creating custom AI Bot personas Feature ai-bot , ai , completed	12	3735	2023 年 11 月 23 日
[Ai Bot] Add user token tracking, custom AI personas, max context posts, document loading, custom API URLs, and localized chat titles Feature ai-bot , ai	2	576	2024 年 3 月 22 日
AI personas integration of the chat plugin: It would be great if site users could chat with AI personas Feature ai , chat , completed	7	1011	2024 年 5 月 11 日

チャット履歴に頼るペルソナのエンジニアリング

関連トピック