Deepset Raises $14M To Help Companies Build NLP Applications

- Advertisement -


Natural Language Processing (NLP), the field of artificial intelligence that involves parsing text for tasks including generalization and generation, is a rapidly growing technology. Based on 2021 data poll of John Snow Labs and Gradient Flow, 60% of tech leaders said their NLP budgets grew by at least 10% compared to 2020, and a third said their spending grew by more than 30%. Fortune Business Intelligence attached $16.53 billion NLP market in 2020.

- Advertisement -

On this background, Deep, the startup behind the open source NLP framework Haystack, today announced it has raised $14 million in a GV-led investment series involving Harpoon Ventures, System.One, Lunar Ventures and Acequia Capital. The capital injection came with Deepset Cloud, a new subscription product for building NLP-based software.

- Advertisement -

“Conditioned [our] Believing in open source, the Deepset team… contributed models and research results to the open source NLP community. [for years]Rusik told TechCrunch via email. “Haystack, the company’s flagship open source product, was born from the experience, knowledge and know-how gained in building NLP for large organizations, and the need for a proper set of building blocks for scalable API-based NLP server applications.”

CEO Milos Rusic co-founded Deepset with Malte Pietsch and Timo Moeller in 2018. Pitsch and Moeller, who have a data background, came from Plista, an advertising startup where they worked on products including an AI ad creation tool.

- Advertisement -

Haystack allows developers to create pipelines for NLP use cases. Originally created for search applications, this framework can provide mechanisms that answer specific questions (eg, “Why are startups moving to Berlin?”) or sifting through documents.

Haystack can also perform knowledge-based searches that look for detailed information on data-heavy websites or internal wikis. Rusik says Haystack has been used to automate risk management workflows in financial companies, returning results for queries such as “What’s the business outlook?” and “How have incomes changed in recent years?” Other organizations such as Alcatel-Lucent Enterprise have used Haystack to run virtual assistants that recommend documents to field technicians.

A stack of hay

Screenshot of the Haystack interface.

According to Rusik, Haystack’s goal was to enable developers and product teams to successfully and quickly build modern API-based NLP applications. He points out that while a data science team can easily build a prototype, there can be challenges going from prototype to production. According to Gartner in 2019, about 80% of AI projects, including NLP projects, will never go into production. poll.

“[With Haystack,] development teams…equipped with all the components to create a fully featured NLP application and guided by proper workflows…Modern NLP is evolving very fast and it is much easier to bridge the gap between cutting-edge research and actual production. off-the-shelf technologies through open source,” Rusic said. “[Prebuilt NLP systems] are the basis [for Haystack] and often give excellent results in pipelines without additional training. Tuning, if needed, happens with end users and experts providing feedback while testing and using new iterations of the product. [system] or pipeline.

But not every company chooses—or wants to—go the DIY route. For those who prefer a managed solution, there is the aforementioned Deepset Cloud, which supports clients throughout the lifecycle of an NLP service. The service starts with experimentation, i.e. testing and evaluating the application, adapting it to the use case and creating a proof of concept, and ends with marking and monitoring the application in production.

“All NLP services that are developed [with Deepset Cloud] can be used in any end application by simply integrating the API,” said Rusik. “Application examples are NLP-based enterprise search (think of ‘Google-style modern search’) and knowledge management.”

With new funding (a total of $15.6 million), Deepset is looking to translate its success with open source — thousands of organizations now use Haystack — into increased revenue. Rusik says the 30-person company, based in Berlin, Germany, was up and running and broke even before raising its first funding round in 2021 and now has large corporate clients, including Airbus.

“[With the new funding,] we will continue to work on the open source Haystack NLP project, adding additional features that will make it even easier for NLP-savvy backend service developers to create NLP services,” said Rusik. “[We’ll also] turn Deepset Cloud into a complete enterprise software as a service for building language applications. This will include creating more agile workflows, more detailed product lifecycle guidance, and offering core and additional tools such as labeling and data integration.”


Credit: techcrunch.com /

- Advertisement -

Stay on top - Get the daily news in your inbox

DMCA / Correction Notice

Recent Articles

Related Stories

Stay on top - Get the daily news in your inbox