Finance automation firm Ramp has acquired artificial intelligence-powered customer support platform Cohere, the company announced last month.

Cohere's use of generative AI and large language models — developed before ChatGPT's arrival — helped to extract companies' customer support data and apply those to similar questions in the future, automatically resolving around 60% of tickets accurately, Ramp said in a statement. Terms of the deal were not disclosed.

"As the AI wave picked up, our team has been noticing the opportunity to augment and improve the workflows and efficiencies of financial teams," said Eric Glyman, co-founder and CEO of Ramp. "We felt this was the right time to be doubling down on just the power of leveraging AI for customers, and this was the most extraordinary team that we have done business with."

Cohere assisted in improving the New York-based fintech's massive sets of structured and unstructured data so customers can avail themselves of the best software price, the company said.

"At Cohere, a lot of our expertise was working with large amounts of unstructured data that we can process and derive enterprise insights and push value towards, and that's kind of the core of our product," Rahul Sengottuvelu co-founder of Cohere, said. "Over time, the product we built was getting these insights out of large data sources from our other customers, including Ramp, and providing value.”

Founded in 2020, Cohere has used generative AI and large language models since 2021 to provide accurate and reliable assistance to customer tickets automatically. The firm has attracted over 200 customers and $3.5 million in venture capital from Initialized Capital and other investors, according to the company.

The acquisition follows the Ramp’s $750 million funding round last year, which highlights its efforts in bringing innovation to the financial services sector, Glyman said. Though the funding allows Ramp to bring in technologically strong teams and "build a stronger company, really the funding was around the business opportunity and strengthening the value proposition," he said.

Ramp, which offers corporate card and expense management services, counts a number of rivals in the business-to-business space, including Brex and Bill. Last year, Ramp said it aimed to further grow its market share by enabling international payments for its business customers.

Ramp was an early customer of Cohere, and adopted large language models to power their customer support.

"It felt like a very different approach. It was taking learnings [from] what our agents were saying and not only improving responses but suggesting and getting to the right answer quicker and it fell transformative," Glyman noted.

Ramp’s recently launched Ramp Intelligence product suite uses AI for various use cases like autocomplete for accounting categories based on learnings from the customer base to automatically code transactions correctly and speed up month-end close.

Also, the fintech’s AI digests contracts on customer’s behalf to detect anomalous language and flag renewal deadline reminders, and its Price Intelligence benchmarks software contract quotes against transaction data — all to give customers the best possible price.

The Cohere team is looking for ways to use AI in a safe and controlled way such that customers can audit the outcome, said Yunyu Lin, co-founder of Cohere. 

AI to save time and money

Around 100 people at Ramp are either building towards, leveraging or working in an AI-related space to offer better customer service, according to Glyman. The fintech aims to make zero-touch expenses so that customers don't need to hold on to the receipt and snap photos.

The co-founder said Ramp's primary focus is to save time and money for its customers. 

"I envision a world in which — just like when you open your Waze app or Google Maps and you know the fastest way to get to a destination — that before renewing, you actually know if there are better prices available," Glyman said. "And we think that the Cohere team and the technology that they have developed is phenomenal truly for taking learnings at scale and information at scale to develop better products that deliver on these goals."