“There’s an app for that.” We all remember this term in 2009 when Apple’s app store famously added more than 100,000 apps within its first year of launch. Thanks to the rapid advances and proliferation of generative AI, today there’s a “co-pilot” for that. Originally coined by Microsoft-owned GitHub for an AI tool that gives coding suggestions, today the term co-pilot is broadly used to describe AI assistant tools that help users improve productivity across a variety of skills and tasks. The recent Cambrian explosion of co-pilots is an important step on the path to next-gen AI assistants — known as “autopilots.”
What is the difference between the two? Put simply, co-pilots help you while you fly, autopilots do the flying for you. Rather than support you in real time while you create content, plan an event, or book travel, an autopilot is more akin to a human assistant — it will complete a task autonomously — freeing you up to do other things. Building an autonomous AI is a marvel of science and engineering, and the team at NinjaTech AI has made it a reality thanks to Chief Science Officer Arash Sadrieh.
Arash first met NinjaTech AI Founder and CEO Babak Pahlavan over 20 years ago in high school, where the two learned to code. Even back then, the concept of AI and autopilot was a burgeoning spark as they competed in national computer tournaments, eventually winning their country’s presidential prize for the most innovative project — a conversational computer that you could talk to. The fuse now lit, Arash accelerated his AI ambitions through academic studies, pursuing his Bachelors in Software Engineering, Operational Research and Systems Analysis and his doctoral degree and post-doc work in Computer Modelling and Simulation. His thesis sought to pave the way for improving the computational performance of process modeling and simulation software applications, demonstrating an order of magnitude improvement in computational time on standard CPU methods.
Prior to joining NinjaTech AI, Arash spent six years at Amazon Web Services (AWS) as a Research and Applied AI Scientist. He accelerated quickly through the ranks, but found himself working on projects that were AI-supported rather than AI-led. As AI technology advanced, and Arash’s passion for building autonomous agents swelled, Babak and Co-founder Sam Naghshineh were simultaneously planning to build NinjaTech AI. After decades, the serendipitous moment had finally arrived for these high school friends to build something truly next-gen. Arash became employee #1 at the company, as its Chief Science Officer.
The team started with the product vision of AI autonomously taking over tasks and worked backwards. Building NinjaTech AI’s autopilot — named Atlas — has been a tremendous exercise in balancing science and engineering, and infusing both with considerable pragmatism. In Arash’s own words, “Science is about solving the initial problem, engineering is about making the solution scalable and extending its efficiencies.” Put differently, scientifically getting an AI to autonomously book a meeting doesn’t equate to it autonomously booking flights for you (and then doing it for 10,000 users) — there is a chasm in between. To achieve a scalable autonomous AI, Arash needed to build something truly novel — an AI agent that could self-learn, self-validate and self-code.
Inspired by DeepMind’s AlphaGo™, NinjaTech AI has gone beyond solving static games and built autonomous AI agents that continuously adapt to dynamic environments. No playbooks and static decision trees, Arash has successfully built agents that adjust using multi-agent game theory and goal-oriented execution to complete complex tasks. NinjaTech’s agents take a step-by-step hierarchical approach to learning, decomposing complex problems into digestible tasks, and creating an execution plan with self-validation. Our AI agents utilize transfer learning to efficiently apply previously acquired skills to rapidly acquire new ones, and they are capable of autonomously writing and validating their own code. This blend of science and engineering means that Atlas is now able to draw on the similarities of booking a meeting with booking a flight, and self-validate in a safe environment until it perfects it.
Since Arash joined Babak and Sam at NinjaTech AI, the team has welcomed over 34 top-tier engineers, scientists and product managers from Google, Meta and AWS. Ready to launch the Beta version of Atlas in early 2024, NinjaTech AI will give a meaningful amount of time back to every busy professional by taking administrative tasks off their hands. If you’re interested in seeing Arash’s science and engineering come to life, try our myninja.ai for free now.
Stay tuned for upcoming product launches and to learn about our next-gen technology in the second founder Spotlight about our VP Go-To-Market — Kurt Wilkinson!
NinjaTech AI is an artificial intelligence company based in Palo Alto, California. The company’s mission is to save every busy professional time and money, by democratizing access to a safe and reliable Personal AI. NinjaTech AI has raised $6 million pre-seed capital from prominent investors such as Candou Ventures, DCVC, SRI Ventures, Jeff Ullman and Laszlo Bock.
