Articul8 AI Inc., a startup that helps companies build artificial intelligence applications, has closed the first tranche of a $70 million funding round.

TechCrunch reported today that the investment was worth over $35 million. Adara Ventures led the deal with participation from chipmaker NXC Corp., Aditya Birla Ventures and others. Articul8 is now valued at more than $500 million.

Santa Clara, California-based Articul8 spun out of Intel Corp. in 2024 with the backing of DigitalBridge, a private equity firm currently being acquired by SoftBank Group Corp. It sells an AI platform derived from technologies developed by the chip giant. Intel uses the software to power an AI assistant that helps its engineers troubleshoot fab equipment. According to Articul8, its platform also lends itself to a range of other use cases.

The software includes a collection of prepackaged AI agents that each focus on a different set of tasks. One analyzes manufacturers’ engineering diagrams to find ways of making production lines more efficient. Another helps utilities optimize grid infrastructure. Articul8 also provides agents for banks, carmakers and other companies.

The newest addition to Articul8’s agent lineup, the Table Understanding Agent, debuted last month. It’s designed to extract information from spreadsheets. According to Articul8, the agent is better than ChatGPT at data interpretation tasks such as determining whether a text snippet is the title of a column or a table.

The agents that ship with Articul8’s platform are based on custom AI models. Companies can use those models to build automation applications. Articul8 says that custom AI software can be deployed on its platform in under ten weeks, which is about four times faster than the rollout times offered by the competition.

A built-in tool called LLM IQ Agent API helps developers identify the most suitable AI model for a new application. Usually, comparing AI models requires software teams to put together custom evaluation datasets, which can take a significant amount of time. Articul8 says the LLM IQ Agent API skips that step.

Some applications use not one but multiple AI models. Articul8’s platform includes a module called ModelMesh that can coordinate the models used by a service, monitor their performance and generate audit logs. The module works with not only the platform’s built-in models but also third-party algorithms.

Articul8 expects to generate $57 million in revenue this year. It currently has 29 customers, including Intel, Amazon Web Services Inc. and other tech industry players, and plans to grow its installed base by investing its newly raised funding in feature development and international expansion initiatives.

The company expects to close its Series B round later this quarter.

Image: Unsplash