Looking back at PearX S19 alum Gradio’s journey, now part of Hugging Face

Intro:

PearX S19 alum Gradio, which is now part of Hugging Face, has had a momentous few years. They recently launched Version 4.0 of their app and they have quickly become a leading workflow tool in the generative AI infrastructure space. We’re so proud of their success, and wanted to take a look back at the earliest days of the company, why we were excited to partner with Gradio’s founders from day 0, and some of their biggest milestones along the way.

How we met the team:

We first met Gradio’s founding team, Abubakar “Abu” Abid, Ali Abdalla, Ali Abid, and Dawood Khan through Pear’s Fellows program. They were housemates at Stanford at the time, and they came to us with an idea to speed up the process of collecting and labeling data for use with AI and ML. Put simply, they wanted to make it really simple for ML engineers to build and share computer vision models and ultimately to make more reliable models. 

Why we invested:

After meeting the team, we were excited to invest in Gradio from day 0 for a few key reasons:

• The team: We knew this was the right team to tackle the space. Abu worked on this problems during his PhD in ML at Stanford. The founding team built deep technical products at companies like Tesla and Google. They also had an amazing advisor in Stanford Professor James Zou, who pioneered data valuation methods. 
• The big market opportunity: Gradio was founded in 2019 when every company was on the precipice of becoming a data-driven or AI company. In 2019 alone, companies were spending $32 billion on data acquisition and labeling, and that number was slated to rise 50% year over year. This easily made this a multi-billion dollar market. 
• The right product vision: We felt that Gradio could solve the biggest problems that data companies face. At that time, to build AI products, companies had to collect and manually label lots of data and then feed that data into machine learning algorithms and there was a crisis of poor data quality. It was a long and broken process that was ripe for innovation. Gradio’s product leveraged ML research to integrate with a company’s existing data pipeline to maximize the value of the data for ML. Essentially, they created the missing data valuation layer to maximize the potential of data for machine learning. 

How they evolved and what’s next:

They joined our PearX S19 cohort, and through the 14 week PearX cohort, they made huge leaps and bounds with their product. They ran four pilots with different kinds of natural language processing (speech or text) companies, ranging from legal contracts to financial records. Over the course of PearX, these smaller pilots led to landing bigger clients like Wells Fargo and TDBank. They used the learnings from this pilot to steadily expand to cover more areas of machine learning like video. 

At the end of PearX, Abu presented at our Demo Day to a room of tier one investors, and after Demo Day, Gradio successfully raised a seed round. Following the fundraise, our team continued working closely with Gradio’s team to find true product-market fit. This included exploring enterprise solutions across various verticals, which eventually led the team to pursue an Open Source approach to expedite product adoption. Gradio was open sourced, and it became the de facto tool for presenting AI / ML projects to a wide range of audiences. In the end, more than 300,000 demos were built using Gradio. 

In late 2021, they were acquired by Hugging Face. The Pear team partnered closely with Gradio’s leadership throughout the entire acquisition process. They are now a key pillar of Hugging Face, where they provide Hugging Face’s users, developers, and data scientists the tools needed to get high level results and create better models and tools. It’s a machine learning match made in heaven and together, they are building the future of ML. 

We’re excited for even more success from Gradio and will be cheering them on!