New Benchmark Lets Businesses Test Language Model Document Handling Skills

Indico Data has launched a platform where businesses can evaluate and share their large language models’ document understanding abilities.

Indico contends that most language model benchmarks focus on “chatbot-related tasks.”

It developed a platform that lets businesses and developers evaluate their AI models on how they extract data from inputs and perform tasks like summarizing and classifying documents.

Benchmark users can employ the test to uncover performance levels and evaluate the running costs of large language models.

Designed for IT executives, data scientists and business decision-makers, users can compare their systems with others through the platform’s leaderboard.

"Indico has been committed to fostering transparency and trust within the AI industry since our founding," said CEO Tom Wilde. "Our latest initiative, the LLM benchmark site, fills a critical gap in the market by offering factual, unbiased information. This platform is essential for enterprises, providing them with the reliable data they need to select solutions that optimally align with their operational requirements"

Before developing the benchmark platform, Indico developed AI solutions for extracting insights from unstructured data sources, including documents, videos, emails and images. 

Related:AI Safety Institute Launches AI Model Safety Testing Tool Platform

Its Data Intelligent Intake solution automates document handling across industries including health care, finance and insurance.

Life insurance provider MetLife, real estate firm Cushman & Wakefield and IT solutions provider Cognizant are among Indico’s customers. 

Founded in 2013, Indico is based in Boston, Massachusetts and has raised $30.4 million in capital. Its backers include Sandbox Insurtech Ventures, Hyperplane and Boston Seed Capital.