Wow, what a success! 💥
Over 200 AI enthusiasts joined us for the next iteration of GenAI Meetup, co-organized with Notion Capital and NVIDIA.
Our CTO Devang, Radu Bozga and Adam Grzywaczewski delivered inspiring talks, sparking engaging discussions on the opportunities and challenges associated with building AI-native products.
Introducing Dynamic LLMs 🥁
Devang highlighted our unique use of Dynamic LLMs within Airspeed. Currently, Static LLMs like ChatGPT are trained on a static snapshot of the world. It is prohibitively expensive to retrain them frequently on the changing state of the world and to adapt them to take advantage of proprietary data.
For LLMs to be truly useful in enterprise settings we need to solve both of these challenges, and that’s where Dynamic LLMs come in. This approach allows us to finetune our models using dynamic customer data, without compromising on quality, speed, and data security. Here at Airspeed we use this technique within our copilot to help accelerate sales teams.
Your reps lean on the Airspeed copilot during a call: it surfaces the right information and collateral at the right moment, so they spend less time hunting for it and more time selling. 🚀
Additionally, John Harding and Adam Grzywaczewski shared the incredible work that is being done at NVIDIA to support the GenAI ecosystem, and Radu Bozga had a very insightful talk on the European AI landscape.
A big thank you to everyone who attended, and special thanks to our co-organizers and partners: Claire Walker, Radu Bozga (Notion Capital) and Rafah El-Khatib, Anastasia Zorina (NVIDIA)
Frequently asked questions
What was the GenAI Meetup Airspeed co-hosted?
An AI meetup co-organized with Notion Capital and NVIDIA that drew over 200 attendees. Airspeed CTO Devang Agrawal, Radu Bozga, and Adam Grzywaczewski spoke on the opportunities and challenges of building AI-native products.
Why are static LLMs a problem for enterprise use?
They are trained on a static snapshot of the world, and retraining them often enough to keep up, or adapting them to take advantage of proprietary data, is prohibitively expensive. That is the gap Airspeed's Dynamic LLMs are built to close.
