Huge Thing
Startup Booster

The aim of the acceleration was to develop NeverBlink's first product, namely the Jelly knowledge graph transmission protocol. Plans focused on increasing the attractiveness of the product by adding new functionalities, further broadening the group of potential users by providing an implementation in the Python language, as well as integrating Jelly with commercial enterprise-class products.

For the first milestone, we have developed a new feature extending the capabilities offered by  Jelly – Jelly-Patch, a format especially useful for change data capture,  incremental backups, replication in high-availability setups, and other practical and commercially relevant scenarios. The work on Jelly-Patch included an implementation in Java (Jelly-JVM) for well-known RDF frameworks such as Apache Jena or Eclipse RDF4J. Then, we have built and deployed a testing server and multiple Raspberry Pi units to construct a dedicated development-and-test infrastructure, enabling more effective engineering processes across the company. Complementing this hardware foundation, we implemented professional DevOps processes and CI/CD pipelines to ensure code quality and automated testing. Finally, we have managed to start work on the explicit goal of milestone 2: a comprehensive, open-source Jelly implementation in Python. 

Moving on to milestone 2, here we finalized our work on pyjelly by releasing a publicly available implementation equipped with an automated test suite. This stage was critical for reaching the wider data science community and involved integrating Jelly with popular knowledge graph libraries, such as rdflib. To further support adoption, we prepared a public, user-friendly, and comprehensive documentation and openly promoted Jelly in open-source communities. This effort involved targeted social media engagement (through posts and comments on services such as LinkedIn and Discord), as well as the publication of a software talk on the SEMANTiCS 2025 SemDev workshop.

Finally, milestone 3 further bridged the gap between open-source innovation and a business-oriented product by developing targeted modules integrating Jelly with commercial enterprise-grade knowledge graph database software (in this case, Neo4j). These integrations were rigorously tested in realistic operational scenarios, allowing the technology to reach Technology Readiness Level 5 (TRL 5) on the startup’s internal infrastructure. The final deliverables of the project centered on long-term growth. Firstly, the Python implementation of pyjelly was further improved and maintained.  Secondly, in order to ensure the protocol’s commercial viability, a cohesive product offer with a clearly defined monetization strategy was developed. Due to multiple market-related factors, it involved consciously moving on from considering Jelly as a separate product and instead focusing on its viability as part of a larger neurosymbolic platform. Final results of the project were presented during a Demo Day in October 2025 to an audience of innovation managers and corporate representatives.