Indico, a provider of enterprise AI for process automation, today announced that it has raised $22 million. The company says the new funding will enable it to double its headcount in 2021, with hires in sales, partnerships, and marketing, while expanding its channel relationships and integration partnerships to broaden its market footprint.
Process discovery and automation is understandably big business. Forrester estimates that robotic process automation (RPA) and related subfields created jobs for 40% of companies in 2019. According to a McKinsey survey, at least a third of activities could be automated in about 60% of occupations, which might be why Market and Markets anticipates the RPA market alone will be worth $493 billion by 2022.
Indico’s platform, which can be deployed in private cloud or on-premises environments or as a managed service, enables customers to automate the intake and analysis of document- and image-based workflows across the insurance, financial services, and health care industries. It ingests PDFs, Word documents, and other unstructured text, images, and documents with built-in support for optical character recognition and more. Post-ingestion, Indico’s technology processes these documents by applying AI models that can be chained together into pipelines to perform data classification, extraction, and comparison for contract audits, customer onboarding, commercial underwriting, financial document analysis, mortgage processing, billing form reviews, and insurance claims analysis.
Using Indico, developers can overlay additional models, like sentiment- and keyword-detecting models, and explore new insights and signals. The platform also provides dashboards for testing and tuning models and exposing errors, as well as for gathering input from subject-matter experts.
“Despite all the advancements using AI and machine learning to create value around structured data, enterprises are not seeing the same benefits and ROI with unstructured content — all the text, images, documents, contracts, and customer interactions that make up more than 80% of data in most organizations,” Indico writes on its website. “Traditional keyword-based approaches — including taxonomies, classifiers, expert systems, and pretrained dictionary-based systems — are simply too complex, too inflexible, and too expensive to maintain.”
Indico claims to use a technique called transfer learning, where a model tailored to one task is used for another, related task, to deploy AI to unstructured content more effectively. Its dozens of custom out-of-the-box models, which were trained on a dataset of over 500 million documents, ostensibly learn to analyze industry-specific data from just 200 training examples. Moreover, they’re as much as 8 times more compute-efficient than traditional machine learning approaches, Indico claims, and can run on one or two graphics processing units.
“By eliminating much of the trial and error typically involved, Indico makes it much more practical to build and deploy custom models to fit specific business needs, without requiring a lot of data science expertise,” the company writes. “[We] built a huge, generalized dataset trained to understand unstructured content. This allows users to train their own, custom models with up to 1,000 times less data than required with other AI and machine learning solutions.”
Jump Capital and Sandbox Ventures co-led Indico’s series B funding round announced today, with participation from Nationwide’s venture capital arm. It brings the company’s total capital raised to $36 million and follows 300% revenue growth in 2020. The round supplements prior funding from 406 Ventures, Osage Venture Partners, Hyperplane Venture Capital, and Boston Seed Capital.
