By MICHAEL CIRCLE
The Data Group News Service
Silicon Valley startup NinjaTech AI on Monday launched a new personal artificial intelligence assistant called Ninja, touting it as an evolution beyond AI assistants to autonomous agents that can plan and execute real-world tasks asynchronously.
The generative AI company said it is leveraging Amazon Web Services' custom machine learning chips Trainium and Inferentia2, as well as Amazon's SageMaker service, to build and scale AI agents capable of handling complex tasks like research and scheduling autonomously.
"Working with AWS's chips has been a game-changer," said NinjaTech AI founder and CEO Babak Pahlavan. "They integrate easily, can elastically scale via SageMaker, natively support large language models while saving up to 80% in costs and are 60% more energy efficient than GPUs."
NinjaTech trained its models using AWS Trainium chips and is serving them via Inferentia2, allowing for faster, lower-cost training and inference compared to traditional hardware.
Pahlavan said the AWS collaboration accelerated the ability to develop an "AI-based planner and action engine" vital for state-of-the-art agents that operate using highly customized large language models modified through techniques like reinforcement learning.
The four initial Ninja agents can conduct web research, schedule meetings, help with coding tasks, draft emails and give advice. Users can access Ninja at myninja.ai and compare its results alongside models from OpenAI, Anthropic and Google.
Annapurna Labs' Gadi Hutt said AWS is "excited to help NinjaTech AI bring autonomous agents to market" in advancing open-source machine learning.
For more on AWS and NinjaTech AI, visit Press release archive (aboutamazon.com).
***
Want to stay in the know about Data Industry news? Subscribe to our weekly newsletter!
Each Monday we will send out the top stories of the week straight to your inbox. Best part? It’s FREE, courtesy The Data Group News Service.
