For centuries, humanity has sought to extend its sensory capabilities through technology. We’ve mapped the visual spectrum into pixels and transformed sound waves into digital waveforms. These triumphs have given machines the ability to see and hear the world as we do. Yet, when it comes to the sense of smell, we remain in the dark. The relationship between a molecule’s chemical structure and its corresponding scent remains an enigma—a frontier science is still attempting to chart. Unlike light and sound, artificial olfaction (AO) defies simple quantification, existing as a deeply complex interplay of chemistry, biology, and perception. At Scentience, we believe this challenge is not a barrier—it’s an invitation.
We envision a world where artificial intelligence and robotics are equipped with the same nuanced sense of smell that living beings rely on to interpret their environments. From detecting volatile chemicals to navigating by scent, the applications of this capability are as profound as they are urgent. Robotics is currently going through its renaissance moment, but they are being developed with a perception stack that primarily includes some permutation of vision, audio, and language processing. The most primitive form of navigation in biology is olfaction, developed far earlier than any other perceptive modality. As robots and AI become an integral part of our daily lives, enabling them to navigate, authenticate, and reason via scent will become a paramount feature not only for robot capability, but also for human protection. True artificial sentience will only be achieved by incorporating all senses with which humans and animals perceive the world.
Our mission is to translate the invisible language of scent into precise, actionable data through the construction of neuromorphic sensing hardware, real-time fusion software, and foundational AO intelligence.
Our near-term journey is built on six foundational pillars:
- Define: There is no mutually agreed upon data standard for the sense of smell. Images are represented as pixels and .PNG files; sounds as frequencies and .WAV files; speech as syllables and text. What is a “scent”? We attempt to define this by working with the broader olfaction community, gathering consensus among industry leaders, and arrive at a data-driven decision. This will be needed to construct training data for olfaction foundation models.
- Discover: With a defined data standard, we can amass libraries of training data. Vision, speech, and audio each have petabytes of open-sourced data that can be downloaded online to train foundation models for robotics. Olfaction does not have this luxury and we aim to balance this asymmetry. To do this, we are constructing a library of compounds over different sensing methodologies enabling us to amount a plethora of training data to train our AI and neuromorphic computers. This compound library will be the largest foundational dataset ever created for olfaction and will become a Scentient product in itself that can be monetized to propel the company forward.
- Develop: Our neuromorphic computers are designed to detect and analyze a spectrum of chemical signatures in real-time not unlike hounds, insects, and single-cell organisms. With a solid training library, Scentient computers are able to detect a vast array of analytes simultaneously. Many before Scentience have attempted this, but none have developed a low-SWaP product that can address a large battery of compounds specifically tuned for real-time edge robotics. Several high-magnitude industry effects ripple from achieving this singular feat. The sale of Scentient computers will be the second primary revenue driver for the company.
- Decode: Through the development of Scentient technology, we can understand the code of olfaction. Our data and products will be used to recursively fine-tune our own olfactory data libraries, in turn fine-tuning our olfactory data models and appreciating Scentient products for our customers. This giant closed-loop training system will create a hive mind that enables unprecedented olfactory perception for AI and robots. Neither hardware nor software can do this alone - ingenuity in both aspects must be fused to materialize the proper outcome. Data and software constructed from this process will constitute a series of Scentient APIs that other AO developers can use as a springboard. This will be the third near-term revenue opportunity.
- Deploy: As the great David Packard once said, “More businesses die from indigestion than starvation”. Startups must focus to survive, and the Company’s sole focus will be exclusively on the robotics industry to enable scent-based navigation to start. Within the broad spectrum of robotics lie factions of aerospace, agriculture, augmented reality, and environmental monitoring - all of which will immediately benefit from first-generation Scentient computers to be released in 2025. Capturing and mastering this isolated market will give us the ammunition we need to assuredly generate revenue and ensure scalable dominance in orthogonal markets in the future.
- Direct: The above five steps will position Scentient as a pioneer and leader in machine olfaction, giving us the opportunity to guide and influence technological development in the field. As a result, key industry, government, and institutional leaders will be established in order to facilitate safe, balanced, and ethical development. We are excited to announce more on this in the coming weeks.
I have spent the last decade working in sensors, AI, and olfaction and have seen the recalcitrance that various industrial titans exhibit toward developing new technologies. The sense of smell for robotics will only be developed by a startup, which compels us to start and advance Sentience with supreme urgency. Below are some of the problems we are solving by equipping autonomous systems with Scentient sensors and AI.
- Humanoid robots - companies building these plan to employ these in homes and warehouses. They should be equipped with the hyper-sensitive smell for enhanced awareness and navigation.
- Autonomous drug detection - less invasive screenings by robots can be achieved schools and geopolitical borders.
- Faster triangulation of explosives - save the dogs, let’s deploy a robot with Scentient sensors at the airport, war zone, or sporting events.
- Advanced home air quality monitoring - let’s prevent the next pandemic virus by spotting allergens and harmful assays at an unprecedented resolution.
- Augmented reality - vision, speech, haptics, and audio are all present in the newest headsets. More immersive experiences can be created with the sense of smell.
- Agricultural off-gassing - provide an additional level of safety on fertilizer and pesticide dosing over different soil types.
- Aircraft engine optimization - assess atmospheric composition at a given altitude to advise on fuel injection parameters.
- Perfumery and food - add a level of quality control to dosing in cosmetics, food, and drink development (many perfumes are still quality controlled by human “master perfumers”).
- Authentication - dogs authenticate their owners and homes through scent, why can’t a robot?
This manifesto is more than a plan; it’s a rallying cry. The sense of smell is the final frontier of machine perception. Our work is cut out for us, but we refuse to let its complexity deter us. Instead, we embrace it as a challenge worthy of human ingenuity, courage, and persistence. We invite visionaries, technologists, and dreamers to join us in this mission. Together, we will not only redefine what machines can perceive but also transform how they interact with the world through AO.
Let’s materialize the invisible. Join us by contributing to the next Industrial Revolution and giving the sense of smell to machines.
Kordel K. France
Founder