Founder's Note - June 12, 2024

Building with Generative AI


As we continue to witness rapid advancements in Generative AI, the recent introduction of Apple Intelligence, focusing on on-device processing among other developments, marks a significant milestone.

This evolution underscores the potential for obsolescence of devices that cannot support sophisticated on-device processing—a domain where companies like Apple are poised to be leaders.

However, Generative AI, as transformative as it is, often requires users to know precisely what they're searching for, especially in consumer use cases. At scale, many users are not familiar with the technology and lack guidance on how it can assist them in areas like product knowledge, health, and other daily needs outside of productivity—areas where the technology is primed to disrupt.

At Spatial Labs, we are pioneering a different approach to GenAI by leveraging modalities like NFC technology and vision capabilities. Through our upcoming platform Continuity Studio, we're building new tools to allow brands and companies to ship products embedded with precise metadata. This enables immediate, post-acquisition touch-points between the products and their audience. We believe this will power new ways of interacting with AI agents and chatbots, allowing users to instantly understand the context of the object they interact with, eliminating the need for user-initiated searches.

This method not only enhances the user experience by removing typical friction points associated with traditional "search" but also paves the way for more intuitive interactions, where the product itself informs its use and care through direct communication with the device.

Mobile devices are also getting more powerful, allowing for new and novel approaches to interface design and hardware-integrated services. I've always foreseen Apple being a leader in this space due to the deep fusion of both hardware and software. However, this new era will present challenges, particularly regarding cross-platform localization and on-device services.

Even high-end Android devices equipped with the latest Qualcomm chips, such as the Snapdragon 8 Gen 3 (which supports lighter localized models), struggle with fine-tuning and optimization due to performance compatibility issues across various OEMs. Apple's complete hardware and software control sets it apart, presenting a challenge that others might struggle to overcome. This could potentially make it easier for developers and new startups wanting to incorporate offline capabilities into their product definitions and use cases.

Recently at WWDC24, Apple also introduced a new form of Private Cloud Compute: a new frontier for AI privacy in the cloud, as well as their new "Adapter" approach to fine-tuning, which are essentially small collections of model weights overlaid onto the common base foundation model.

Furthermore, the concept of model distillation or context pruning presents a significant opportunity for new startups looking to make their mark in the GenAI landscape via the mobile domain. These methodologies involve refining complex AI models to be more efficient and focused, enabling them to operate effectively on less powerful devices or with less data. For consumer applications, where quick and precise functionality is crucial, the ability to distill AI capabilities into simpler, more efficient forms can be a game-changer.

The true potential of these techniques lies in their ability to tailor AI's broad capabilities to specific consumer needs. In a world where technology increasingly anticipates and responds to individual preferences, startups that can harness these techniques to prune unnecessary context and enhance relevant features will lead the next wave of innovations. This approach not only improves user experience but also drives down costs and technological barriers, making advanced AI tools more widely accessible.

We've been meditating on new ways to embed distilled AI models and chatbots directly into products. This will ensure that each product acts not just as a tool but as a smart, context-aware guide that enriches the user’s experience from the very first touch. This strategy is about more than just technological innovation; it's about creating a seamless, intuitive experience that meets consumers at their point of need—a turning point in user interaction where every product begins to tell a story.

To further this vision, we are exploring innovative uses of computer vision and object classification technologies alongside NFC. By integrating these capabilities, we can offer even more robust AI-driven interactions directly through physical products. This allows our products to become not only items of use but also intelligent agents capable of understanding and adapting to user behavior and environments without cumbersome user inputs. The potential to revolutionize consumer interactions in this manner is immense, as it opens up new avenues for autonomous, context-aware user support and engagement.

As GenAI continues to evolve, the emphasis on practical, consumer-focused applications will undoubtedly grow. We are excited to be at the forefront of this movement, ensuring our users enjoy truly frictionless and intuitive interactions with our products.

We look forward to seeing how this competition unfolds and how our innovative approaches redefine user interactions.

In summary, it’s an exciting time to be in AI. Whether you’re a builder, user, or simply an observer, our world will look much different than it does today in just a few years. I'm looking forward to seeing how this space unfolds.


Iddris Sandu
Founder & CEO, Spatial Labs