AI code discovery platform CatalyzeX raises $1.64M

California-based CatalyzeX, a startup that offers a platform AI/ML code discovery and know-how, today announced it has raised $1.64 million in a seed round of funding led by Unshackled Ventures, Darling Ventures, Kepler Ventures, On Deck, Abstraction Capital, Unpopular Ventures, and Basecamp Fund. The company said it plans to use the round — which also saw the participation of multiple angels — to further accelerate the development of its product, democratizing AI for builders worldwide.

Over the years, tens of thousands of AI researches have been conducted, building a huge repository of technical material for various use-cases and industries. However, finding relevant information from this huge chunk for a project at hand has long been a challenge for developers and data scientists around the world. They’d often end up spending hours on Google, searching papers that could contain code snippets and models to build on (only 10 to 12% share code) and other know-how that could accelerate the development of their AI project.

CatalyzeX for AI code discovery

Above: CatalyzeX AI code discovery platform

Image Credit: CatalyzeX

Prompted by this challenge in their own professional careers, brothers Gaurav and Himanshu Ragtah decided to start CatalyzeX in 2019. The startup offers a website that curates AI research papers and studies from the web, giving devs a one-stop-shop to discover ML techniques and know-how, along with the corresponding code, for their respective projects.

“CatalyzeX’s offering is powered by crawlers, aggregators, and classifiers we’ve built in-house to automatically go through technical papers as well as code platforms daily and to match and link machine learning models and techniques with various corresponding code implementations,” Gaurav told Venturebeat in an email. “We also allow code submissions and feedback from members of the CatalyzeX network.”

The free-to-access platform is a search engine of sorts, where a developer picks the recommendations or puts in a problem query, like cancer detection, in the search field. The results show all relevant available ML models/techniques — with full paper and code implementation — that could help with the problem. If the code is not publicly available, the platform also provides an option to get in touch with the authors to request it or get further questions answered.

In addition to this, CatalyzeX also offers a browser extension that automatically displays links to code implementations for ML techniques and papers appearing in Google Search results.

“Since code is the lingua franca for builders and makers, not walls of text, and given the sheer volume of developments in AI research every single day, surfacing relevant code implementations greatly saves time and effort for developers and technical non-experts in discovering and assessing viable options to leverage artificial intelligence in their products and processes,” the cofounder added.

Focus on addressing current status quo, growing user base

While platforms like 42papers and Deepai.org also offer AI research and know-how, CatalyzeX claims to differentiate with a much larger repository for model/techniques and code discovery. The platform currently serves over 30,000 users every week with more than 500,000 code implementations.

However, Gaurav emphasized that the real challenge is not to beat these sites but to address the current status quo, which is heavily fragmented and holding back significant technology development from reaching the real world.

This, he said, will be done through accelerating the development of the product, taking it to more developers and data scientists around the world. Gaurav did not share specific product development plans, but he did note that a part of the funding will go toward hiring product designers and engineers who would work upgrading the platform.

“We also have integrations and partnerships planned with several code-collaboration and AI research platforms,” he added while noting that they are also exploring monetization options such as introducing a paid tier with advanced search filters and integrations with development/deployment environments or connecting high-skill talent with global opportunities in AI.

According to PwC, AI could contribute up to $15.7 trillion to the global economy in 2030. Of this, $6.6 trillion is likely to come from increased productivity and $9.1 trillion from consumption-side effects.