Contextual AI raises $80 million for model-enhancing technique

By Kenrick Cai

SAN FRANCISCO (Reuters) - Contextual AI, a startup that sells a tool to improve the performance of artificial intelligence models, raised $80 million through a Series A funding, the company told Reuters.

The Mountain View, California company declined to disclose its post-money valuation, but data provider PitchBook pegged the figure at an estimated $609 million.

Led by venture capital firm Greycroft, funds came from a slew of existing investors including Bain Capital Ventures and Lightspeed, which previously pumped $20 million to launch the company in 2023.

Contextual AI's CEO Douwe Kiela previously supervised a team at Meta that invented a technique known as retrieval augmented generation (RAG).

After OpenAI’s ChatGPT propelled generative AI into the mainstream last year, AI practitioners increasingly adopted RAG to help overcome the limitations of off-the-shelf AI models, such as hallucinations which produce inaccurate responses.

RAG works by feeding curated information into an AI model to give it more context so that the output is more accurate and up to date.

"It’s been interesting to see RAG become a completely mainstream idea," Kiela told Reuters. "We don't have to explain it anymore."

The product is not yet generally available. The new funds are focused towards getting it into the market, though Kiela said he did not have a launch timeline.

For now, Contextual AI is selectively working with companies in the finance, technology and media sectors. HSBC and Qualcomm are clients.

Cloud providers like Microsoft Azure and Amazon Web Services and some startups are already offering RAG built into their AI software suites.

(Reporting by Kenrick Cai in San Francisco; Editing by Sonali Paul)