For many, ChatGPT and the generative AI hype train signals the arrival of artificial intelligence into the mainstream. But while there’s little question of a seismic sea change these past six months in terms of public awareness, the growing demand for AI could be outpacing the infrastructure required to power the myriad use cases that are emerging — and this is something that German startup Qdrant is looking to address.

Founded out of Berlin in 2021, Qdrant is targeting AI software developers with an open source vector search engine and database for unstructured data, which is an integral part of AI application development particularly as it relates to using real-time data that hasn’t been categorized or labeled.

To help bring its technology deeper into the commercial sphere, Qdrant today announced a $7.5 million seed financing from lead investor Unusual Ventures, with participation from 42cap, IBB Ventures and a handful of angel backers, including Cloudera co-founder Amr Awadallah. This is in addition to the €2 million ($2.2 million) in pre-seed funding Qdrant raised last year.

Unstructured

Vector databases, for the uninitiated, are geared toward storing such unstructured data, like images, videos and text, allowing people (and systems) to search unlabeled content, which is particularly important for extending the use cases of large language models (LLMs) such as GPT-4 (which powers ChatGPT).

According to Gartner, unstructured data constitutes as much as 90% of new data generated in the enterprise, and is growing three times faster than the structured equivalent. At the same time, the vast majority of AI research and development (R&D) projects never make it into production, which Qdrant CEO and co-founder Andre Zayarni reckons is due to a lack of the right tools — ultimately, being able to connect an LLM to real-time, unstructured data can open up a wealth of opportunities to anyone looking to build more useful AI applications.

“Vector databases are the natural extension of their (LLMs) capabilities,” Zayarni explained to TechCrunch. “The biggest limitation of GPT is that it ‘knows’ only about events that happened before the time the model was trained, but if it’s connected to a vector database, the virtual ‘memory’ of an LLM can be extended with real-time and real-world data.”

Investors have been taking note, too. Just last year, a similar proposition to Qdrant called Pinecone nabbed $28 million, though Zayarni considers Qdrant’s open source foundation as a major selling point for would-be customers.

“Engineers trust open source, and it will be hard for proprietary software to compete in this market if there is an OSS product with a similar — or even better — offering,” Zayarni said.

There are other open source players out there already, of course, such as Zilliz, a startup commercializing the Milvus open source vector database and which raised $60 million last year. And earlier this month Chroma secured $18 million in seed funding to grow its “AI-native” open source vector database.

That Qdrant has now raised $7.5 million in seed funding is somewhat telling about where investors’ heads are right now — any technology that promises to help advance AI and machine learning, and extend its capabilities to all developers, is clearly an attractive proposition.

Zayarni said that Qdrant spent the better part of a month fine-tuning its pitch-deck for its seed funding round, and received its first term sheet on the second day after sending out its deck, which was followed by another term sheet two days after that.

“We had more than 20 VCs interested — almost all of them wanted to join as co-investors later — and we most probably would have received more offers,” Zayarni said. “But Unusual Ventures’ deep experience with OSS (open source software) and its model of being an active operational partner instead of just an investor were incredibly attractive to us, so we decided to go with them.”

Today’s funding news comes a couple of months after Qdrant launched its managed cloud offering, which is designed to help developers through one-click deployments, automated version upgrades, backups and a soon-to-launch database admin interface.

With its fresh cash injection, Zayarni said Qdrant is also working on an enterprise product that can be hosted on-premises or in a private cloud, which he expects to launch later this year.

Qdrant, an open source vector database startup, wants to help AI developers leverage unstructured data by Paul Sawers originally published on TechCrunch