18:34:32 From Tony White to Everyone:
	General Scripts per each Script Step
	https://community.claris.com/en/s/question/0D7Vy000001gfMLKAY/detail
18:37:08 From Ian Jempson to Everyone:
	Specifically how to use them with locally hosted LLMs
18:37:09 From Chris Moyer to Everyone:
	Here's a late-breaking news bit you may not have in the deck: https://www.vox.com/future-perfect/2024/5/17/24158403/openai-resignations-ai-safety-ilya-sutskever-jan-leike-artificial-intelligence?utm_source=bensbites&utm_medium=referral&utm_campaign=openai-faces-safety-questions-as-the-superalignment-team-disbands
18:43:31 From Gianluca D Aquino to Everyone:
	Yep he is developing a reverse proxy for FM Server
18:46:08 From FM Bug to Everyone:
	https://filemakerhacks.com/2024/05/13/onwindowtransaction-json/
18:49:43 From Tony White to Everyone:
	Got to go and prep for my 2pm. Carry on! ;-)
18:57:50 From Gianluca D Aquino to Everyone:
	it has some potential, in particular for final user that will use natural language for complex search, but what about performance?
18:58:01 From FM Bug to Everyone:
	https://acolyteapplications.net/blog/adding-natural-language-search-to-your-claris-filemaker-applications/?et_fb=1&PageSpeed=off
18:58:02 From Gianluca D Aquino to Everyone:
	is it using gpt3.5?
19:04:01 From Sam H. to Everyone:
	It is using gpt-3.5-turbo by default, but you could change it if you need.
19:04:55 From FM Bug to Everyone:
	Thanks for the clarification @Sam H.
19:06:38 From Sam H. to Everyone:
	@FM Bug You're welcome.
19:06:39 From Sam H. to Everyone:
	[This is an encrypted message]
19:10:26 From Ian Jempson to FM Bug(Direct Message):
	Hey Jul, sorry if I’m dense but I don’t get what I’m watching. I understand  it’s doing stuff with DAPI, but not clear on why!
19:14:21 From Gianluca D Aquino to Everyone:
	Reacted to "It is using gpt-3.5-..." with 👍
19:26:24 From FM Bug to Cris Ippolite(Direct Message):
	Hi Cris. Good to see you here. Just finishing off a presentation about building web apps with a FileMaker back end in Five.
19:26:50 From Cris Ippolite to FM Bug(Direct Message):
	👋 sounds good
19:31:52 From FM Bug to Everyone:
	Five.co
19:31:52 From Heidi to Everyone:
	Great podcast (am here to heckle, but also give thumbs ups!)
19:32:30 From Heidi to Everyone:
	https://claristalk.buzzsprout.com
20:00:54 From Heidi to Everyone:
	and then add video and image data as well
20:00:59 From Heidi to Everyone:
	setup for ian :-)
20:01:11 From Heidi to Everyone:
	Cris also has videos on Youtube
20:01:20 From Gianluca D Aquino to Everyone:
	Reacted to "Cris also has videos..." with 👍
20:01:38 From Heidi to Everyone:
	https://www.youtube.com/@iSolutionsAI
20:06:17 From Gianluca D Aquino to Everyone:
	for me is ok
20:09:05 From FM Bug to Chris Moyer(Direct Message):
	Have you managed to see Ian’s screen yet. On my system, it’s in another window to the other attendees and the current speaker.
20:11:06 From Chris Moyer to FM Bug(Direct Message):
	Interesting recent Stack Overflow post about doing embedding of confidential documents on-prem with spaCy: https://spacy.io
20:11:21 From Chris Moyer to FM Bug(Direct Message):
	https://stackoverflow.com/questions/77936280/how-to-create-embedding-vectors-on-premise-for-openai-models
20:20:17 From Heidi to Everyone:
	I run ollama in docker with a WebUI container and an API access container which is really easy way to start up and use in different ways
20:20:37 From Heidi to Everyone:
	there’s a model on there for relationship advice
20:20:43 From Gianluca D Aquino to Everyone:
	Reacted to "there’s a model on t..." with 👍
20:20:55 From Heidi to Everyone:
	https://ollama.com/library/samantha-mistral
20:22:38 From FM Bug to Chris Moyer(Direct Message):
	Replying to "Interesting recent S..."
	
	Thanks Chris
20:23:01 From Heidi to Everyone:
	there’s also an A/B testing of diff. models in LMStudio..
20:23:11 From Heidi to Everyone:
	if you have the latest lmstudio
20:37:53 From Chris Moyer to FM Bug(Direct Message):
	https://ollama.com/library/phi3
20:38:01 From Chris Moyer to FM Bug(Direct Message):
	"Inappropriate or Offensive Content: these models may produce other types of inappropriate or offensive content, which may make it inappropriate to deploy for sensitive contexts without additional mitigations that are specific to the use case.
	"
20:46:43 From domferrari to Everyone:
	Very interesting Ian we are going to have to sign off
20:48:02 From Sam H. to Everyone:
	Which vector DB would you suggest to get start?
20:48:35 From FM Bug to Everyone:
	Replying to "Very interesting Ian..."
	
	Thanks for coming!
20:51:09 From Chris Moyer to FM Bug(Direct Message):
	https://nirantk.com/writing/pgvector-vs-qdrant/
20:57:10 From Carl Henshall to Everyone:
	Got to go but thanks for the presentations
20:57:49 From Chris Moyer to FM Bug(Direct Message):
	https://huggingface.co/blog/embedding-quantization#binary-quantization-in-vector-databases
20:58:07 From Chris Moyer to FM Bug(Direct Message):
	How you store the embedding affects performance
20:58:41 From Chris Moyer to FM Bug(Direct Message):
	Must dash. Great meeting!
21:01:41 From Sam H. to Everyone:
	Thank you very much
21:30:43 From Ian Jempson to Everyone:
	http://www.humanizingwork.com/wp-content/uploads/2020/10/HW-Story-Splitting-Flowchart.pdf
21:31:20 From Sam H. to Everyone:
	Reacted to "http://www.humaniz..." with 👍