Akridata

Akridata is a Los Altos, California company that develops software for AI-powered visual inspection and for managing the visual data used to train computer vision models. Founded in 2018, the company initially focused on the problem of visual data at the edge, launching the Akridata Edge Data Platform in October 2021 — described by the company as an edge-core-cloud platform that creates and manages smart data pipelines and AI workflows, filtering, curating, tracking and staging large streams of video and lidar data generated by distributed edge devices. The platform was positioned in what the company called the "Data-Centric AI" category, supporting traceability of data from inception to inference and complementing explainable AI efforts.

The current product line consists of three offerings: Vision Assist, a programmable image-capturing system that augments manual quality control with automated defect detection and real-time analysis, including 2-axis and 3-axis controlled inspection stations and vertical inspection systems for large parts; Vision Command, a real-time command center for defect detection, compliance tracking and production optimization; and Vision Copilot, a platform for data science teams to build, refine and deploy computer vision models with data curation, labeling, analysis and synthetic data generation. A related Data Explorer product lets teams connect to visual data in place across S3, Azure and Google Cloud Storage or local sources, cluster and embed datasets to find outliers and biases, run image, patch, natural-language text-to-image and video-sequence search, evaluate model accuracy, view saliency maps, and visualize or generate classification, detection and segmentation metadata.

Akridata markets to manufacturing and production teams, quality and compliance managers, automation and industrial engineering teams, and AI/ML and data science teams, across manufacturing, automotive, medical devices, railway, critical infrastructure, food and agriculture. The company's website carries a banner stating that Akridata has been acquired by Dimaag-AI.

Founding story

Akridata was formed in 2018 by a team of serial entrepreneurs — described as Silicon Valley entrepreneurs, data scientists and product engineers who collectively hold nearly 200 patents — with the aim of solving the data problem in the AI training lifecycle, specifically around visual data at the edge. The name derives from "Akri", Greek for "edge". Kumar Ganapathy is co-founder and CEO. A Cornell startup listing names Sanjay Pichaiah (MBA '07) as a founder.

Business model

Akridata licenses software products — Vision Assist, Vision Command, Vision Copilot and Data Explorer — to enterprise customers, with self-service trial entry points ("Try Data Explorer", "Try Vision Assist") and demo booking, and a pricing page on its site. Products are designed to integrate into customers' existing legacy inspection systems and to access customer data in place across cloud object stores or local sources rather than requiring data migration.

Traction

By the October 2021 platform launch the company had more than 50 team members. Customers cited include Toyota Motor Company North America, whose Director of Engineering described using Akridata to automate data pipelines from edge locations to the cloud for ADAS/AV development. The company states its platform has processed multiple tens of petabytes of visual data since 2018, and reports platform benchmarks of ten times faster time-to-access relevant data, four times more efficient compute and storage usage, and two times better data scientist and ML engineer productivity. A published medical device customer case study reports a 40% reduction in defect rates plus automated compliance tracking.

Latest developments

Akridata's website carries a banner announcing that the company has been acquired by Dimaag-AI. Recent published resources cover surface defect detection with AI in steel and metal industries and traceability in medical device sterilization QA.