Hi All, Open sourced a quick setup for an agentic engineering team setup,
One 1000 line script that spawns - AI developers agents with varying enginnering roles that mimic a software development team. Its Unlicense, Have fun!!! ✨🦄
https://t.co/YJAgayrihw
Boston OpenClaw 🦞 November meetup was packed with some great demos and fun conversations
1. Pablo Gonzalez Rodriguez: inspired the room with Exo (his personal AI OS), a custom harness working with over 30 agents to orchestrate his comms, learning, and other aspects of his life
2. @allnash: presented EggRun, a "The Safe Computing framework" with modern AI-based code harnesses, infrastructure, and software.
3. Pavel Tkachyk: demystified preserving and isolating context in agent memory with Substrate, both for private and enterprise use.
4. Hema Seshadri, Ph.D., John Zinky, and the Northeastern University team: showcased Multi-Agent Brainstorming Assistant to Choose Vacation Destination
We also had a surprise visitor from the @OpenClaw Foundation, @Pishgar, who leads developer relations and communities at the Foundation.
We want to thank Harvard Business School AI Club for hosting us at the HBS campus and sharing their space with us. Also thankful to my cohosts Albert Feng-En Tsai, Wade Kimbrough, and Austin Chen Wang.
Boston OpenClaw Meetups are monthly events, and we invite partnership and collaboration. If you want to present a personal AI project with the community, DM me.
Please join our Boston OpenClaw Discord server to engage with local community members. https://t.co/7DefnEws03
#BostonOpenClaw
@ashbhatia Safe Computing is becoming increasingly important. #OpenClaw is a solid tool and in its #own machine its even more powerful and #utilitarian.
Set one up in two clicks.
https://t.co/WFuPRgRvNj
Doing electronics surgery and hoping to not get kicked of the cafe. ☕️Hoping to showcase this #Startek styled tricoder. 🔥🕹️✨ using Egg. 🥚
Getting ready for #PhysicalAI summit next week to showcase @EggRunAI@johnkwerner@imaginationxyz
Whether you are a human or an agent -- read your training logs.
I was playing with @karpathy's autoresearch for quite a bit already and noticed that the agent has very limited observability into the training process and rarely looks beyond the final validation loss.
I updated `https://t.co/qMX4Cv03pr` to log more training statistics and added an analysis step where the agent uses Python to inspect training dynamics. This little change makes the autoresearch agent achieve ~60% better validation loss.
@karpathy Exactly the same pain. With many years of education in CS, it feels like I had no skills to ship complete end-to-end apps.
Here's a course where we attempt to solve this pain: https://t.co/0tRuCTf1BX
Its our first attempts and covers the basics, but still there's so much more.
Hi All, Open sourced a quick setup for an agentic engineering team setup,
One 1000 line script that spawns - AI developers agents with varying enginnering roles that mimic a software development team. Its Unlicense, Have fun!!! ✨🦄
https://t.co/YJAgayrihw