Jina AI Raises $30M to Advance Neural Search

Jina.ai Secures $30 Million Series A Funding for Neural Search Technology

Berlin-based Jina.ai, an open-source company specializing in neural search, has announced a $30 million Series A funding round. Canaan Partners led the investment, with participation from Mango Capital, GGV Capital, SAP.iO, and Yunqi Partners.

Advancing Beyond Keyword-Based Search

This funding brings the total raised by Jina.ai to $39 million. CEO and co-founder Han Xiao, alongside Nan Wang and Bing He, established the company with the goal of surpassing the limitations of conventional keyword-based search methods.

The core of Jina.ai’s innovation lies in utilizing deep learning neural networks. This approach leverages technologies like transfer learning and representation learning to empower developers to construct tailored search solutions.

How Neural Search Works

“We employ deep neural networks to convert various data types – images, audio, video, and more – into a universal representation,” Xiao explained. “This representation is primarily a mathematical vector, typically 100-dimensional.”

“Instead of counting matching letters, the matching algorithm calculates the mathematical distance between these vectors. This methodology can address a wide range of data search and relevance challenges.”

Jina as the TensorFlow of Search

Xiao positions Jina as analogous to TensorFlow, Google’s open-source machine learning framework. Just as TensorFlow and PyTorch established patterns for AI system design, Jina aims to become the standard for building neural search systems.

Expanding the Jina Ecosystem

Beyond the core Jina framework, the company offers Jina Hub, a marketplace for developers to share and discover components for Jina-based applications.

They also recently launched Finetuner, a tool designed for fine-tuning deep neural networks.

Building a Comprehensive Development Experience

“We’ve dedicated the last 18 months to constructing the foundational infrastructure for our neural search platform,” Xiao stated. “Now, we are focused on developing a complete end-to-end development experience.”

The Jina AI developer community currently comprises approximately 1,000 users.

Real-World Applications

Applications of Jina’s technology are diverse. A video game developer utilizes it to automatically populate game assets within their editor, while a legal-tech startup employs it to enhance a chatbot’s Q&A capabilities using data from PDF documents.

The Power of Open Source

Since its launch in May 2020, the open-source Jina framework has attracted nearly 200 external contributors. The company also maintains an active Slack community.

“We embrace open source due to its inherent velocity, which I believe is crucial for software success,” Xiao emphasized. “Open source fosters rapid development and iteration, vital for infrastructure software.”

“We are building a community to gather feedback and improve usability and accessibility quickly.”

Future Plans

Jina.ai intends to use the new funding to double its team size and expand its North American operations. Investment will also focus on research and development to broaden the Jina ecosystem and introduce new tools and services.

Industry Perspective

“Traditional search systems struggle with multimedia data,” noted Joydeep Bhattacharyya of Canaan Partners. “Jina AI is transforming how companies unlock unstructured data, offering a fast, scalable, and data-agnostic solution.”

“The early applications of its open-source framework demonstrate the potential of neural search to improve decision-making, streamline operations, and generate new revenue streams.”