00:01 OpenAI Assistants API V2 brings cost and latency improvements 01:52 OpenAI Assistants API v2 Release brings new features for better context management 05:34 Improved connection to documents and efficient chunking of files 07:25 RAG is a middle point solution for efficient context token usage 11:02 OpenAI Assistants API v2 offers cost-effective model fine-tuning 12:42 OpenAI's E store works with naive chunking strategy and multihop document reasoning 16:25 OpenAI Assistants API v2 brings efficiency and cost reduction 18:07 AI Assistants API v2 allows filtering results from different knowledge bases 21:39 AI Developer discussing support for CSV and implications for data extraction 23:38 OpenAI Assistants API v2 Release 26:58 OpenAI Assistant API v2 has improved speed and cost-effectiveness. 28:48 OpenAI's commanding lead will grow with the release of version 4.5 or 5.
This is awesome news! Thank you, as always, for sharing this update and your expertise. We WILL work together one day, as soon as our firm is large enough to build our first SAAS.
As I am building the knowledge base of a custom gpt that I have been working on I would love it if they would allow chatgpt to update the knowledgebase for me. As it stands right now I need to update the files offline, then re-upload them every time I want a change. I find myself wishing as the creator of the gpt that I could just say it in a conversation, to remove, or add sections of knowledge to a file in the knowledgebase. It would make it so much easier.
It is best if OpenAI supply multiple different base model for fine tuning for different purpose like medical, business, science, art, etc. The base model will a lot smaller and hopefully faster interference.
Interested in exploring what AI can do for your business? Get in touch with the Morningside AI team here: morningside.ai/
Liam, I love the way you are like a filter for nerd speak, so it's easier to grasp for regulars.
Thanks for all you do. 🙏
00:01 OpenAI Assistants API V2 brings cost and latency improvements
01:52 OpenAI Assistants API v2 Release brings new features for better context management
05:34 Improved connection to documents and efficient chunking of files
07:25 RAG is a middle point solution for efficient context token usage
11:02 OpenAI Assistants API v2 offers cost-effective model fine-tuning
12:42 OpenAI's E store works with naive chunking strategy and multihop document reasoning
16:25 OpenAI Assistants API v2 brings efficiency and cost reduction
18:07 AI Assistants API v2 allows filtering results from different knowledge bases
21:39 AI Developer discussing support for CSV and implications for data extraction
23:38 OpenAI Assistants API v2 Release
26:58 OpenAI Assistant API v2 has improved speed and cost-effectiveness.
28:48 OpenAI's commanding lead will grow with the release of version 4.5 or 5.
Thanks, this is some of the best coverage of these updates that I was able to find.
Appreciate you all so much! Great video!
What are your thoughts on LangGraph approach? How would you compare to Assistant API v2?
This is awesome news! Thank you, as always, for sharing this update and your expertise. We WILL work together one day, as soon as our firm is large enough to build our first SAAS.
@morningsideAI is the cost of a conversation lower with openai assistant api in comparison to chat completion api
Thank you Liam , can you please make another video that show how to connect this assistant to Voiceflow ? Please.
Great, informative video!
Demo please...
Please make video that shows how to connect this V2 to voiceflow with Replit ?
As I am building the knowledge base of a custom gpt that I have been working on I would love it if they would allow chatgpt to update the knowledgebase for me. As it stands right now I need to update the files offline, then re-upload them every time I want a change. I find myself wishing as the creator of the gpt that I could just say it in a conversation, to remove, or add sections of knowledge to a file in the knowledgebase. It would make it so much easier.
Interested in your teams preferred agent framework? Or do you do all custom?
It is best if OpenAI supply multiple different base model for fine tuning for different purpose like medical, business, science, art, etc. The base model will a lot smaller and hopefully faster interference.
So it's almost as good as what we can build ourselves in a couple days
I love AI ❤
I want to watch this, but the “follow the bouncing ball” text is too much.