Snowy Persian New Year - Weekly Update, Mar. 25, 2024

MORE FROM ME: Substack | YouTube | Events | Book me to Talk | Course Survey

Last Week at Aggregate Intellect!

This past week we had an important demo with one of our clients. The goal was to demonstrate that the LLM system we have been POC'ing with them can be deployed and they can use it on their own. There were a lot of nuances that we needed to get right, importantly deciding what is absolutely necessary for the demo and what can wait a bit longer to be added / fixed. I thought the demo went pretty well, and we did get some early positive feedback, along with a few things that we need to prioritize fixing.

I also spent a big chunk of the week with prospective customers, explaining our framework and engagement model, and understanding their use case and how we can help them. A few of these conversations are getting very close to the finish line, so I'm very excited about reporting back on how they turned out next week!

If you remember from last week, I mentioned that I'm designing a course to be offered on Maven in cohort-based way. The course will focus on the product aspects of multi-agent LLM systems. This week I analyzed the data from the survey I did and used that to design the landing page of the course, as well as the outline. Still some work to do to even get to a meaningful syllabus, but getting pretty close. I have a deadline to wrap it up this week, so stay tuned for the page coming online next week.

Unsung Hero(s)

I have been putting together a sales pitch deck to communicate the value proposition of the work we do. Lemuel Amen ,Steve Luengo and Bryan Lehman spent some time with me giving feedback about the deck which arguably made it much more convincing. Thank you, Lem, Steve and Bryan.

Abi Aryan ?? made an intro to Rod Rivera who signed up as the final speaker of our upcoming workshop, and now you can see the full line up for May 3rd here. Thank you, Abi and Rod.

LLM Stuff

• I asked Alireza Darbehani about LLM Fine-tuning vs Training
• I asked Andy McMahon about long-context LLMs and RAG
• With Suhas Pai , Nikhil Varghese , and Ian?? Yu we talked about ways to improve a RAG pipeline
• I asked Jay Alammar about the relationship between language and intelligence
• I saw some interesting work on using "evolution" principles in merging models to automatically create multi-modal foundational models

Startup Stuff

• I didn't write as much this week! But promise! I'll make it up next week
• I celebrated Persian New Year (started on 23:06ET on Tuesday March 19th) with some fellow founders and AI folks by walking through snowy trails of mid-town Toronto, and some Persian food!

Upcoming Events

• LLM Problem Solvers - Competitive Advantage in Era of LLMs
• LLM Problem Solvers - LLM RAG Chunking Strategies
• May 3rd, One day workshop on LLM products and user experience

Podcasts

• Seeing Ourselves through Dark Moods | Mariana Alessandri | Talks at Google | book
• Kevin Kelly - Be Generous and Unique
• How to fund real change in your community

Call to Action

BOOK ME TO TALK

RESPOND TO MY COURSE SURVEY

READ MY ESSAYS ON SUBSTACK

SUBSCRIBE TO OUR YOUTUBE

SUBSCRIBE TO OUR EVENT CALENDAR