The company is announcing $22M in Series B funding, bringing its total raised to over $40M in less than 2 years since the company’s founding.
In addition, Abacus.AI has deconstructed its autonomous AI platform into 3 stand-alone modules that can be used by organizations to bring AI models to production.
Abacus.AI is announcing $22M in Series B funding led by Coatue. Decibel Ventures and Index Partners also participated in this round. As part of this release, the company is announcing Abacus.AI Deconstructed, which is a suite of 3 stand-alone tools that can be used by organizations to bring AI models to production.
Series B funding
With this round the company has raised $40.3M in total funding in less than two years. Yanda Erlich, General Partner at Coatue, is joining Abacus.AI’s board of directors. “We are proud to be leading the Series B investment in Abacus.AI,” said Erlich, “because we think that Abacus.AI’s unique cloud service now makes state-of-the-art AI easily accessible for organizations of all sizes, including start-ups. Abacus.AI’s end-to-end autonomous AI service, powered by their Neural Architecture Search invention, helps organizations with no ML expertise easily deploy deep learning systems in production.”
Mike Volpi from Index Ventures and Jon Sakoda from Decibel also participated in this round. “We are excited to continue to invest in Abacus.AI and support their mission to democratize AI and make state-of-the-art deep learning systems available in a plug and play fashion to organizations of all sizes,” said Volpi.
Read More : More Brands Are Prioritizing Programmatic Advertising to Meet the Increasing Consumer Demand
“Every executive can imagine how AI can be used in our business operations, but few have been able to capitalize on its benefits because of the cost and technical resources required. Abacus is working to democratize AI by offering a turnkey solution for the most widely needed use cases, and brings the promise of deep learning into any application without complexity or cost,” said Sakoda.
Abacus.AI Deconstructed
Abacus.AI offers a state-of-art autonomous deep learning service that automates all aspects of machine learning, from model creation to deployment and maintenance. As part of this release, the company is announcing Abacus.AI Deconstructed. Deconstructed separates parts of Abacus.AI’s underlying platform and offers them as stand-alone modules. Today, when less than 1% of models trained are actually put into production, these 3 services will help organizations quickly deploy and maintain models in production.
Model Hosting and Monitoring – Organizations can easily host their models in production with this module. This module helps teams deploy, maintain and govern them all in one place that will assist with the messy issues of production operations. Monitoring models for drift is essential for maintaining ML models in predictions and knowing when to trigger re-training runs. This module alerts production teams when models experience prediction and data drifts.
Model Explainability and DeBiasing – One of the biggest barriers to AI adoption is the black-box nature of model predictions. With this module, organizations get explanations for each of their predictions and can dig into why a particular prediction is unexpected or counterintuitive. The beauty of the Abacus.AI service is that it helps organizations easily determine if there is bias based on age, gender, or race in the AI model and applies debiasing techniques that they have previously open-sourced. The debiasing module is based on a research paper that Abacus.AI will be presenting at this year’s NeurIPs conference.
Read More : How Data Analytics and Content Strategy Can Drive Effective Marketing Campaigns
Real-time machine learning feature store – This service allows organizations to easily create, store and share machine learning features and deploy real-time machine learning and deep learning models models in production. By creating and maintaining integrity across both online and offline features and checking pointing dataset and feature versions, it is easy for organizations to deploy real-time machine and deep learning systems into production.
