Physical AI; New Thermal Insulation Material for Space and Defense; Machine Learning System-on-Chip; Plant-based pigments for Cosmetics, Fashion, Paint, and Packaging markets
Physical AI; New Thermal Insulation Material for Space and Defense; Machine Learning System-on-Chip; Plant-based pigments for Cosmetics, Fashion, Paint, and Packaging markets
An insider’s update on Deep Tech: Your dose of tech innovations, startups, exponential industries, policies, and market moves to stay ahead and capitalize on it.
Physical AI; New Thermal Insulation Material for Space and Defense; Machine Learning System-on-Chip; Plant-based pigments for Cosmetics, Fashion, Paint, and Packaging markets
Welcome to The Scenarionist’s Sunday Brief, where we discuss 5 crucial facts from the week shaping the Global Deep Tech landscape at the intersection of technology, policy, and markets.
Get practical insights from cutting-edge tech innovations to the most significant startups and capital moves to stay ahead, stay competitive, and seize new market opportunities.
Subscribe now to get our briefs delivered straight to your inbox - For Free.
In today's briefing, we delve from AI applications in the physical domain to novel thermal insulation materials poised to revolutionize space and defense sectors. We also explore cutting-edge machine learning system-on-chip technology that promises to redefine edge computing, alongside pioneering efforts to introduce plant-based pigments into the cosmetics, fashion, paint, and packaging markets. These developments not only highlight the dynamic nature of the deep tech landscape but also underscore the profound impact these technologies are poised to have on our everyday lives and the global economy.
Physical AI: How AI Can Transform Physical World Insights
What if AI could help us solve real-life problems in the physical world? What if it could help us to discover hidden patterns and behavior we couldn’t see before? - says Archetype AI on LinkedIn
Archetype AI, a pioneering startup based in Palo Alto, California, this week emerged from stealth mode and announced a $13M seed funding backing from Venrock, Amazon Industrial Innovation Fund, Hitachi Ventures, Buckley Ventures, Plug and Play Tech Center, Earl Grey Capital, and Strange Ventures, as well as several angel investors.
Large Language Models (LLMs) have long been at the forefront of AI research, showcasing impressive abilities in generating text, images, and even videos. Despite their advances, these models face a crucial limitation: their reliance on a narrow range of input data types, which starkly contrasts with the inherently multimodal nature of human perception. Humans integrate a broad spectrum of sensory data—text, speech, visuals, touch—to understand and navigate the world. Recognizing this gap, Archetype AI introduces Newton™, a model that signifies a leap towards multimodality in AI, capable of deciphering the world in a truly physical, sensory manner.
This innovation transcends the digital constraints of traditional LLMs, harnessing a multitude of data from environmental sensors—cameras, radars, accelerometers, and more—to provide insights into the physical world in real-time. This approach not only amplifies the model's utility across various sectors but also democratizes access to sophisticated sensor data interpretation. By converting complex multimodal inputs into human-friendly outputs—text, visuals, and even code—Newton™ facilitates an intuitive interaction between humans and the physical environment, from industrial operations to everyday life scenarios.
“We put sensors in all kinds of things to help us, but sensor data is too difficult to interpret. There’s a potential to use AI to understand that sensor data—then we can finally understand these problems and solve them.” asserts Brandon Barbello, a co-founder and COO of Archetype AI.
The essence of Archetype AI's innovation lies in its novel multimodal foundation model, which integrates sensor data in multiple dimensions and modalities, understanding it in a contextual and temporal framework. This large behavioral model (LBM) excels in detecting, classifying, and predicting objects and human behavior by merging physical data with AI's cognitive capabilities.
The potential applications of this technology are vast, promising transformative solutions in robotics, automotive, consumer devices, healthcare, insurance, retail, and, obviously, defense and security.
In the top frame, the person is stationary, and this status is reflected on the rightmost display stating "A person is stationary," while the waveform and spectrogram show minimal activity, indicating little to no movement. In the middle frame, the person has begun to move slowly, and the technology captures this change. The text display updates to "A person is receding slowly," while the waveform monitor shows slight fluctuations, and the spectrogram's colors begin to shift, signaling a change in sensor input. In the bottom frame, the person's pace has increased. The devices now register more significant activity and The text output adjusts accordingly to "A person is receding quickly.
One of the most compelling aspects of Newton™ is its capacity to democratize access to complex sensor data, enabling businesses and developers to ask open-ended questions about their operational environments and make informed decisions based on real-time insights. This represents a significant departure from traditional data analysis methods, which often require specialized knowledge and are limited by the complexity of the data itself.
“Newton fuses multimodal temporal data – including signals from accelerometers, gyroscopes, radars, cameras, microphones, thermometers, and other environmental sensors – with natural language to unlock insights about the physical world in real-time.” asserts Ethan Batraski, Partner at Venrock on LinkedIn
However, the path to realizing the full potential of physical AI is fraught with challenges. The integration and interpretation of multimodal data necessitate advanced algorithms and models capable of understanding the nuances of the physical world, in addition to managing regulatory and privacy concerns.
Moreover, the conversation around foundational models, as highlighted by Matt Wood of AWS, at an AWS analyst meetup, points to a future where AI models become increasingly specialized and tailored to specific industry needs. This specialization is both a challenge and an opportunity, as it requires significant investment but also offers the promise of highly customized solutions that can address specific industry challenges.
However, innovations like these today are the gateway to a new era of physical intelligence, where the boundaries between the digital and the physical blur, enabling a more intuitive, effective, and insightful interaction with the world around us. The implications for industry are profound, offering the promise of not only solving existing challenges but also unlocking new opportunities for innovation and growth.
Hi there! The Deep Tech Briefing is a product dedicated to
It's a weekly brief that provides updates, market signals, key innovations, deals, and essential information startup leaders should be aware of in the Deep Tech world.
Not yet part of this community? Sign up today to read the entire brief and gain access to all the exclusive content!
