Meet Spice AI: Empowering Developers with AI and Data Infrastructure 

Protocol Labs
Protocol Labs
Meet Spice AI: Empowering Developers with AI and Data Infrastructure

Snapshot: With Spice AI (opens new window), developers no longer need to be data, artificial intelligence or machine learning experts to engineer software. By providing core building blocks for creating data and AI-driven products, including compute, storage, zero knowledge and machine learning accelerators, and blockchains, Spice AI supports developers within one interconnected application for their building needs.

The Impact: The creation of AI-driven applications with seamless access to web3 data has immense implications. The platform simplifies the intricacies of data infrastructure, enabling developers to build and ship products more easily. By removing the complexities of working within web3 tools as they evolve, Spice AI’s streamlined approach means less pain for developers, lower costs, and less labor.

On Joining the PL Community: In 2021, Spice AI secured pre-seed funding with contributions from Protocol Labs. Led by founders Luke Kim (opens new window) and Phillip LeBlanc (opens new window), both with deep roots in developing tools for developers at Microsoft and GitHub, the partnership with PL was catalyzed by a shared vision for the future of web3 infrastructure.

What's on the Radar: Going forward, Spice AI is hyper-focused on deploying (opens new window) OSS use-cases, enabling developers and teams to query and deliver data to their applications, dashboards, and ML pipelines without trading off between cost and performance.

# The Genesis of Spice AI

Spice AI, founded by Luke Kim and Phillip LeBlanc in 2021, aims to solve a significant challenge: bridging the AI application development gap for developers beyond the realms of tech giants like Google and Microsoft. The founders were motivated by the insight that many developers lack the tools and expertise to integrate crucial AI-driven decision-making. “It became clear to me while working at Microsoft that decision-making AI is going to be critical to many applications going forward,” Kim said. “I also realized that there's really no platform out there for developers that helps you build this type of software.”

AI-driven decision-making in software development involves leveraging artificial intelligence to automate and enhance decision processes within applications. This technology enables software to analyze data, learn from it, and make or suggest decisions without human intervention. Applications range from optimizing logistics to personalizing user experiences on digital platforms. AI decision-making empowers software to adapt and respond dynamically, leading to more efficient operations and improved user satisfaction.

To bridge this gap, Spice AI launched a platform (opens new window) designed to democratize AI development, making it accessible and straightforward. By offering developers a suite of tools and resources to incorporate AI into their projects, Spice AI levels the playing field for developers and paves the way for the next generation of intelligent, data-driven applications.

“So with Spice AI, we set out to build a set of building blocks. Simple composable, incrementally adoptable building blocks that help developers build AI-driven applications without having to be experts in them,” Kim said. “When we spoke with a company, they said, ‘Yeah, we want to do AI, we know we have to, we just don’t have the expertise. There’s no platform to do it.’ And so our goal is to build this platform that helps those types of developers actually build applications and be able to compete in the future.”

# Tackling Challenges Head-On

While initially focusing on its open-source platform to direct its mission, the team quickly recognized the critical need for high-quality, real-world data to power AI models. Data stands at the core of building effective AI tools, including large language models (LLMs). This realization accelerated a focus on blockchain data, overcoming one of the most significant hurdles in AI development. "Even if you have a good API, if you don't have enough high-quality data, it's very, very hard to develop the technology," Kim said.

The significance of blockchain data lies in its openness, diversity, and the volume of real-world transactions it captures, making it an invaluable resource for developing next-generation AI-driven applications. This unique dataset makes an excellent resource for training sophisticated AI models capable of nuanced understanding and decision-making. “Now, if you think about big leaps in AI in human history, they've all come from really two things. One is much faster, much cheaper hardware. And so you see that now with GPUs and with computers and Moore's Law. But the other thing that's really given us step changes in AI is bigger and better high-quality datasets,” Kim said.

Recognizing the scarcity of high-quality open data necessary for building intelligent models, the team worked on developing blockchain indexes and enabling high-performance SQL queries on blockchain data. This not only addressed significant parts of the data challenge, but also attracted Spice AI’s first paying customer, marking a pivotal moment in its growth. In 2022, they successfully secured $13.5 million in additional funding to further fuel their development efforts. This infusion of capital marked a significant phase in their expansion, underscoring investor confidence in their approach to leveraging blockchain data for AI model training.

# Charting the Future

As Spice AI transitions from a promising start-up to a key player in the AI and data infrastructure domain, it is refocusing on simplifying AI integration even further, ensuring developers have more accessible, efficient, and powerful tools at their disposal. “We currently have a cloud platform of managed data and AI services that people can use, but what we’re excited about is improving what we think of as the last mile of data delivery,” Kim said.

The emphasis on optimizing for 'last mile' data delivery highlights Spice AI's commitment to advancing AI and data integration for developers. He expects that soon, developers will be able to run a distributed node next to an application or in a specific cluster, enabling developers to deliver data directly to that node and run AI models locally. “So instead of having to rely on a single central service to do inferencing to build an AI-driven application, you can now run that much closer to your application in a distributed manner,” Kim said.

In March 2024, the team also completely redesigned OSS using Rust (opens new window), creating a unified SQL query interface and a portable runtime. This enables the local materialization, acceleration, and querying of datasets sourced from any database, data warehouse, or data lake. This means developers can spend less time wrangling data and more time creating innovative applications and business value. Spice AI encourages the developer community to contribute at (opens new window).

Spice AI’s efforts are gaining traction. Recent use cases include acceleration and co-located datasets with applications and frontends, to serve more concurrent queries and users with faster page loads and data updates, as exemplified by the CQRS sample app (opens new window). Another use case is the streamlined approach to monitoring queries via the Sales BI Dashboard. (opens new window)

“If we're building a developer focused platform that helps you build AI decision-making applications for next generation applications, what is going to be next when hardware is continuing to get faster and cheaper? What is going to be the next massive open dataset that drives another step change in AI development?” said Kim. “We believe that it's going to be with free and open blockchain data.”