Startup One AI Inc. today launched from stealth mode with $8 million in funding to help companies add natural language processing features to their applications.

The funding was provided by a group of backers that included technology investor Ariel Maislos, Tech Aviv and SentinelOne Inc. Chief Executive Officer Tomer Wiengarten. Several other unnamed venture capital firms and angel investors participated as well, TechCrunch reported.

Natural language processing technology has advanced significantly in recent years. In 2020, OpenAI LLC debuted an artificial intelligence model dubbed GPT-3 that is optimized for natural language processing and can perform tasks ranging from writing essays to producing software code. In April, Google LLC detailed an AI system dubbed PaLM that it says is even more advanced than GPT-3.

In the enterprise, natural language systems can potentially be used to automate a wide variety of manual tasks. In practice, however, the technology is often difficult for organizations to implement. That’s the challenge One AI is working to address.

Many natural language processing models are available under an open-source license, which means that companies can in theory easily add them to their software. But before a neural network can be deployed, it has to be trained, a process that requires significant time and effort. It often also requires a significant amount of technology infrastructure.

Some natural language processing models are available in pre-trained versions. However, the pre-trained version of a neural network isn’t necessarily suitable for the task that a company is hoping to automate. A neural network trained to process academic papers, for example, may not be capable of efficiently processing business documents and vice versa.

One AI provides a collection of pre-packaged natural language processing models optimized for business use cases. They’re available through an application programming interface, which enables developers to integrate the startup’s technology into their software with relative ease.  

The company’s pre-packaged neural networks can sort documents by topic, extract customer sentiment from help desk tickets and create text summaries. The startup also supports a variety of more specialized use cases. Developers could, for example, use One AI’s software to create an AI application that detects when a website visitor who is interacting with a chatbot may require assistance from a help desk representative.

Developers can use One AI’s pre-packaged neural networks separately or deploy them side-by-side. A company could combine multiple AI models to create a software workflow that sorts customer support requests by topic, creates a summary of each request and then highlights important keywords in the text. To ease application development for customers, One AI offers a tool called One AI Language Studio that enables developers to mix and match neural networks without writing code. 

The company has 22 employees, who include 10 natural language processing researchers. It reportedly plans to use the new funding round to hire more people.

One AI’s launch from stealth comes a few weeks after Accern Corp., another startup working to make natural language processing easier to implement, raised $20 million in funding. Accern uses natural language processing to help financial firms find useful insights in their business data. Companies use the startup’s technology for tasks such as tracking stock market developments and evaluating investment opportunities. 

Photo: Unsplash