How Klarity Accelerated AI Adoption with a ‘Generative AI Sabbatical’

To figure out what was possible for Klarity with large language models that could understand and produce natural language or code, Nadhamuni placed a team of engineers on a “generative artificial intelligence sabbatical.”

For four weeks, employees put all other new development work on hold and focused on how to use these designs to improve Klarity’s product and delight customers. The only rule was radical ideas only, nothing incremental.

“I basically abdicated all other responsibilities and just worked on AI, 12 hours a day, every day for a month,” said Nadhamuni, who studied computer science and machine learning at MIT. “We just took our product apart and rebuilt it using generative AI.”

The result: Klarity unlocked new levels of automation for processes that were previously performed by humans—increasing the pass rate to 85 percent—and accelerated its ability to deliver features.

Investors took notice. In June, Klarity raised $70 million in a Series B round led by Friedman and Gross, with participation from Scale Venture Partners, Tola Capital, Picus Capital, Invus Capital and Y Combinator. The round brings Klarity’s funding to more than $90 million. Has not shared a rating.

“Of our portfolio, they were by far the fastest to move,” said Aaron Fleishman, a Klarity investor through his early-stage venture firm Tola Capital.

It’s a sink or swim moment for tech companies. Just as the Internet transformed industry by industry 40 years ago, the rapid and unstoppable march of AI is changing every corner of society. For startups, it means technology adoption is no longer optional; it is quickly becoming the lifeline that companies need to stay afloat in a competitive market.

“Their window is closing,” Fleishman said of companies that are still not adapting. “Either they can quickly become a near-native AI player in the market, or they will fail because they will be disrupted.”

“The rules of physics have changed”

Even before his sabbatical, Klarity was well-versed in natural language processing and computer vision. Its software has turned documents like order forms, invoices and other contracts into structured data that machines can understand as well as humans. This helped accountants save time and avoid costly mistakes.

Early versions of its product were built on a multitude of third-party models, open-source libraries, and models that Klarity trained itself. But Nadhamuni said his own models were limited in their ability to process documents they had never seen before because they were trained on small data sets. When OpenAI released the GPT-3.5 large language model, it opened up new possibilities for technology firms using their own custom models.

Klarity’s sabbatical began with a three-day executive team meeting to get them on the same page. Nadhamuni gave a talk on generative AI, its history and the latest developments in the technology. On the first night, he gave the executives homework: play with ChatGPT on personal and work tasks.

“The way I framed it, the rules of physics changed,” Nadhamuni said. “Before we think, we need to know what the new rules are to really understand what might be possible.”

On the second day, the executives huddled around a conference room table and threw a Miro board on a screen. They wondered how generative AI could transform the way their product extracts data from documents and understands the nuances and context of the data. They whittled hundreds of suggestions down to five key hypotheses. If the new technology could do these five things, it would rebuild the core of the Klarity platform using generative AI.

For example, a total reset is worth considering if the new technology was better at understanding shapes and data it hadn’t seen before. If the new technology made integration easier for non-experts, it would completely change the way Klarity sells to enterprises.

Engineers broke into small groups to address each hypothesis. Nadhamuni said employees have “complete autonomy to do what they want, use the tools they want, obviously respecting data governance issues.” They met on Friday to share a summary of the past week’s learning.

Nadhamuni has also written a manual on how to run a generative AI sabbatical, which is available on the Tola Capital blog.

By the end of the four weeks, Klarity was ready to push the nuclear button on its custom models in favor of using large language models. The GPUs he used are gathering dust in a desk.

According to Nadhamuni, a sabbatical allowed the company to move at the speed of rapid technology development. It shortened software development cycles from weeks and days to hours. And a sabbatical forced the team to radically rethink what was possible.

“What I always tell my team is to assume there are 17 iterations between where we are today and the right answer,” he said. “Don’t obsess over the right answer, obsess over how to repeat it quickly and in an informed way.”