'With generative AI, spatial storytelling is at an inflection point'

Sravanth Aluru, Co-founder, Avataar
 
The world of commerce is being overhauled to go 3D, Aluru, who is also the CEO of the 3D artificial intelligence (AI) technology platform Avataar, tells Forbes India in a recent interview. With $55 million in funding from investors including Tiger Global Management and Sequoia Capital, Avataar’s team of about 180 specialists in Silicon Valley and Bengaluru serve customers ranging from Sleep Number in the US to Pepperfry and Bajaj Auto in India. Edited excerpts:

Q. You mentioned that the industry has leapfrogged over the last 18 months or so. Give us a sense of what you’re seeing.
From a 10,000-feet perspective, Avataar is a generative AI platform focused on spatial storytelling. Spatial storytelling emphasizes the depth of 3D reality, bridging the gap between digital and physical worlds. It reimagines how consumers engage with digital content through photorealistic elements, blending digital objects seamlessly into the physical environment for a compelling experience.

In our focus on industry trends, two pillars stand out: Generative AI for cost-effective content and experience creation, addressing the 3D industry's need for scalability. The shift towards spatial computing with Apple Vision Pro validates our vision of a 3D world bridging digital and physical realms.

AI advancements, like transformers and neural radiance fields, bolster confidence in 2D to 3D inference. The platform shift may initially target enterprise applications, allowing users to explore digital inventory in physical stores, while widespread consumer adoption could take five to ten years. We remain adaptable, supporting various devices on the path to that future.

Q. Tell us about what you’re doing with generative AI
In the realm of generative AI, like ChatGPT, I observe its broad reach, with applications focusing on text and voice. The next frontier involves images and videos, where diffusion models excel in predicting RGB pixels.

 At Avataar, we take it a step further, extending generative AI to the 3D world. In this context, understanding goes beyond RGB pixels and delves into the true laws of physics, encompassing lighting, material properties, reflections, and more. Disentangling RGB pixels into underlying physical parameters forms a crucial part of our AI engine, aided by neural radiance fields.

 This complexity allows us to grasp the world at a more granular level, giving us an edge in applied AI use cases, especially in the 3D domain.

Also read: AI, robots will never replace humans but can make us more productive and efficient: Thomas Frey

Q. Give us a couple of examples of what your customers are able to do with Avataar today that they were not able to do, say 18 months ago or even a year ago
In our work at Avataar, we focus on two main areas of impact. The first is amplifying impact across channels, spanning the upper, middle, and bottom funnel. By leveraging spatial storytelling in various formats, like 3D ads or D2C activations, we've achieved significant conversion uplifts, as seen with Sleep Number's 6.7 percent increase, after we went live with them.
 
This approach aligns with consumers' desire for more informed purchase decisions. The scale of evidence is growing, with transactional outcomes now becoming evident, not just engagement levels. The second area centres on generative AI cost efficiencies. Historically, creating 3D content incurred high costs due to manual artist-driven processes.

However, machine learning advancements have brought efficacy, leading to self-learning intelligence that predicts and learns better with more data. We've implemented reinforcement learning and AI safeguards to ensure accurate representation without hallucinations, safeguarding consumer trust. As we scale up and gather more data, new emergent capabilities are likely to emerge, making our platform even more powerful.

Q. So from here, can you talk about how you see Avataar evolving?
In our ‘Avataar Inside’ strategy, we've developed foundational generative AI intelligence capable of understanding and representing the world in 3D. Initially, we focused on commerce, using photorealism as a core pillar to make digital and physical reality indistinguishable. Now, we see the potential to apply this technology in various areas, such as gaming, real estate, and industrial applications.

We're exploring opportunities within commerce, like sampling and prototyping, as well as virtual photography, which streamlines photo shoots using 3D representations. The depth in commerce lies in uncovering opportunities for spatial depth to enhance consumer experiences. Additionally, the breadth of application extends to adjacent markets like gaming and real estate.

To facilitate this growth, we've opened up our SDK to enterprise customers and individual developers worldwide, leveraging the power of spatial computing and generative AI. The community response has been inspiring, and we're excited about the possibilities that lie ahead.

Q. Tell us about your partnerships and business expansion.
Our goal is to be present wherever consumers are. We focus on existing supply-demand ecosystems where 2D commerce prevails and aim to bring the power of spatial depth to enhance consumer experiences. In the upper funnel, we partner with companies like Meta and Google to drive brand awareness and concentration, providing an alternative to traditional 2D formats with higher return on investment (ROI).

In the middle funnel, we tap into direct-to-consumer opportunities by collaborating with agencies and platforms like Shopify and BigCommerce. This allows brands to easily integrate our technology and upgrade their stores to 3D. In the marketplace segment, our partnerships enable brands to list products on major platforms like Samsung, upgraded to 3D through our technology.

 Partnerships are a fundamental part of our growth strategy, facilitating integration, co-innovation, and expansion into existing ecosystems for maximum impact. We continue to forge alliances with AI-focused companies like Nvidia and cloud providers such as Google Cloud and Azure, leveraging co-innovation opportunities for our platform's advancement.

Also read: The promise and peril of artificial intelligence as a general purpose technology

Q. Give us a sense of how you’re growing
In both enterprise and platform partnerships, we serve large marketplaces across various commerce categories, upgrading high ASP and high traffic products to 3D experiences. Our rendering SDK enhances the front-end consumer experience, leading to an immediate 100 percent uplift in engagement, while interactive feature narratives can increase impact by up to 3.5 times.

The growth is driven by ROI and AI's efficiency in automating tasks, benefiting both artists and clients. As cost barriers decrease, the catalogue’s 3D penetration will increase over the next few years, serving all customers and expanding our reach across marketplaces.

We are experiencing strong tailwinds in the AI and generative AI space, with a notable mindset shift among clients towards embracing AI's capabilities and focusing on its 90 percent success rather than dwelling on the 10 percent failures. Budget allocations for AI-driven cost efficiencies have increased, and the ChatGPT and LLM inflection have driven this change.

With millions of impressions across various formats, we continuously experiment with new visual formats and reimagine consumer experiences. While we don't yet have metrics on headsets, the impact is evident on desktops, mobile web, and in-app contexts, driving significant engagement and transactions.

Q. Does the SaaS playbook work for you as well – with development teams in India?
Our lean team in San Francisco comprises a couple of PhDs from academic institutions like Berkeley AI Lab, focusing on connecting various academic research dots. The team acts as a bridge between ML experts and researchers, leveraging the evolving research to enhance our product's capabilities.

 Most of our intellectual horsepower is based in Bengaluru, benefitting from the strong talent base and math skills of Indian engineers. We have around 180 team members. An important senior addition is Dan Firth, formerly from Ikea, leading our 3D platform from London.

We raised $45 million in early 2022, but we haven't utilized the funds extensively due to our strong (low) burn model. Fundraising is currently opportunistic, not a survival need. We're in a build mode, focused on value creation. While we started with a core focus on the US, we now see AR penetrating global markets across different geographies. We're expanding our reach to Europe, Asia, and the Middle East, aiming to become a global destination for our platform.