Return to search

Khosla Ventures and CEO Benioff back MetaMind

MetaMind, a machine learning company in Palo Alto, Calif., has secured $8 million in funding. The investors were Khosla Ventures and Marc Benioff, CEO and chairman of


PALO ALTO, CA—December 5, 2014 — Today, MetaMind launched its state-of-the-art artificial intelligence (AI) platform for enabling automated predictions and smarter decisions with greater accuracy and speed than ever before. The company has developed ground-breaking deep learning software, combining natural language processing and computer vision, to solve a broad range of complex data problems.

MetaMind’s breakthrough technology is based on co-founder and company CTO Richard Socher’s deep learning research. Deep learning is a rapidly growing branch of AI and machine learning that comprises a set of techniques that doesn’t require domain experts to program knowledge into algorithms. Socher’s work has been recognized by peers and leaders in the field as a major leap forward in improving accuracy in natural language processing and computer vision. MetaMind’s technology outperforms many other systems on several public benchmarks.

“Textual and image data are growing exponentially, and is already at a scale where human beings are simply not capable of processing it effectively. Recent and rapid advances in machine learning in general, and deep learning in particular, are making possible solutions to problems that couldn’t be addressed before,” said Sven Strohband, MetaMind CEO and an operating partner and CTO at Khosla Ventures. “We are pushing the performance on several established benchmarks in vision and language understanding, and making the resulting state-of-the-art technology available to business users via our AI platform.”

MetaMind’s general purpose AI platform can easily be applied to solving a variety of real-world problems, such as classifying cancer in mammograms; finding anomalies in X-ray radiology images; identifying food from individual images and presenting relevant health and nutrition information for each item; automatically answer customer emails on behalf of a company that has millions of users and high volumes of customer inquiries; accurately identifying sentiment for financial texts and 2014 ImageNet Competition. 2014 SemEval Competition Task 1, Stanford Sentiment Treebank, Microsoft Research Paraphrase Corpus, ETH Food-101 dataset assisting in algorithmic financial trading; automated social media analyses; and automatically calculate the fair price of a vehicle from processing images of the vehicle. MetaMind is currently working with Fortune 500 companies across industries to implement smarter, faster and more accurate decision-making solutions.

The company also announced initial funding of $8M from Salesforce Chairman and CEO Marc Benioff and Khosla Ventures.

“MetaMind has developed some of the most impressive technology I have ever seen,” said Benioff. “Richard, Sven and team are doing amazing work and already delivering impressive results with real customers. The company’s deep learning technology is going to have enormous impact in multiple industries, and has the potential to provide a generalized mathematical model for building machine intelligence that could be adapted for almost any discipline.”

“I am very impressed by the depth of skills MetaMind has assembled, and its fast progress to match and even outperform the field on machine learning benchmarks to date,” said Vinod Khosla, founder of Khosla Ventures. “The team’s breakthrough work on deep learning algorithms and architecture is remarkable, and I’m excited by the ongoing brainstorming into the myriad of possible use cases that could make machine learning broadly accessible to high-value business users.”

MetaMind Smart Modules
MetaMind offers businesses a state-of-the-art set of Smart Modules, which convert unstructured or semi-structured information, such as pictures and sentences, into mathematical objects from which they can make decisions with little or no supervision from humans.

Example Smart Modules are publicly available on the MetaMind Labs homepage, They include:
● General Object Classifier: Classify images into thousands of visual categories
● Food Classifier: Snap an image food and it identifies what it is
● IcMe – Image Classification Made Easy: Train your own classifier on any problem via “Drag, Drop & Learn” in the Web browser
● Textual Similarity: Identify and match similar strings of text based onsemantic similarity
● Sentiment Analysis: Identify positive, neutral or negative sentiment in a text, or social media like Twitter
● etcML: Text Classification: Train your own classifier on any problem via “Drag, Drop & Learn” in the Web browser
MetaMind is based in Palo Alto, CA, and currently has 10 employees. The company’s advisors include Chris Manning, Professor of Computer Science and Linguistics at Stanford University; Yoshua Bengio, Professor in the Department of Computer Science and Operations Research as Université de Montréal; and Richard Boyle, an operating partner at Khosla Ventures.

For more information on MetaMind, or to start leveraging deep learning for your business decisions, please visit:

About MetaMind
MetaMind is a machine learning company based in Palo Alto, CA. Founded in 2014, the company provides simple solutions for complex business problems, leveraging AI to tap into large volumes of data in order to make automated predictions and smarter decisions with more speed and accuracy than ever before. For more information, please visit: (going online on Dec 5th)