Advex’s generative AI platform reportedly reduces the time required to develop production grade AI vision models by drastically compressing the data collection process.
SAN FRANCISCO—Advex AI, a startup tackling a critical data problem in AI Vision for manufacturing, recently announced its $3.5 million seed funding round, led by Construct Capital. Also participating in the round were Pear VC, Emerson Collective, and angel investors Arash Ferdowsi (Dropbox founding CTO) and Ankit Jain (Infinitus CEO), Advex AI said in a release.
“We chose Advex for PearX, our accelerator for early-stage founders, because they showed remarkable technical expertise and a clear vision for tackling one of AI’s toughest challenges,” said Arash Afrakhteh, partner at Pear VC, in the release. “From day one, they’ve proven themselves as top-tier founders with an innovative approach.”
Advex also reported that it was selected as a top 20 Finalist in the 2024 TechCrunch Disrupt Startup Battlefield.
Challenging the traditional AI workflow
Advex’s platform is reported to represent a radical shift in how enterprises develop production-grade AI Vision models. Instead of undergoing months of expensive data collection cycles, AI and automation teams can reduce the time to hours by leveraging Advex’s GenAI platform. By doing so, they can automatically figure out what data is missing, then synthetically create that missing data using advanced diffusion models, the company said.
“By combining data collected from the real world with use-case specific synthetic data, Advex ensures that AI Vision models are trained on comprehensive and balanced datasets, dramatically improving AI’s performance and reliability in real-world environments,” the release stated. “And when the real-world changes, Advex can quickly generate relevant synthetic data and keep AI Vision models at production-level accuracy.”
According to Advex AI, manufacturers and companies in other critical industries can now implement and maintain production-grade AI vision solutions in days rather than months. This addresses the urgent need for automation to drive speed, effectiveness, and efficiency amidst a growing labor shortage.
“AI adoption is no longer optional for manufacturers, yet the lack of usable data creates ROI barriers that make most automation projects uninvestable,” said Dayna Grayson, co-founder and general partner at Construct Capital, in the release. “Advex fundamentally changes this dynamic, providing immediate value in automating manufacturers’ processes and turning months-long projects into days. We believe they’ll become the foundation for how the global manufacturing industry automates.”
“We’re tackling a major problem—the labor shortage—by unlocking fast and high-ROI automation in the world’s most fundamental industries,” said Advex Co-Founder and CEO Pedro Pachuca, in the release. Pachuca previously researched computer vision with Google Brain and led a consumer electronics startup.
Seven of the world’s largest manufacturers already use Advex AI’s vision technology, according to the company. More than 25 additional billion-dollar companies are reported to be currently adopting Advex’s capabilities.
“Customers routinely benefit from a 50 percent-plus improvement in visual automation performance in just a few days, a feat which would normally take many months, if not years, to achieve,” the company said in the release. “automotive, aerospace, and consumer electronics industries are strong adopters, leveraging Advex’s technology for a multitude of use cases including surface defect detection and vision-guided robotics.”
Advex, an NVIDIA Inception program member, also partners with leading machine vision providers to quickly reach customers and integrate deeply into customer workflows.
By focusing on the industrial sector, Advex said it can amass real and synthetic industrial data that is impossible to find on the internet. This, combined with real-world production feedback, reportedly enables Advex to “develop the best synthetic data and AI models for industrial applications.”
To support rapidly increasing demand from global manufacturers, Advex AI is actively hiring across various roles.
“Data is the biggest problem in all of AI—if you have the right data, you have AI,” said Advex Co-Founder and CTO Qasim Wani, in the release. “In the past, I’ve worked on AI for chip design and drug discovery applications. And the number one reason why it took forever to productionize AI models was how quickly I could collect high quality data. By solving data, you solve AI.”
Wani previously worked on deep learning with Qualcomm and PathAI and “built products used over 50 million times, the release said.