Ireland-based Zerve AI Ltd., a data science and artificial intelligence development platform, said today it raised $7.6 million in a seed round to help data scientists and engineers collaborate faster.
The company's seed round was led by Paladin Capital Group, with participation from existing investor Elkstone and angels including Rob Hickey, former executive vice president of DataRobot. Zerve previously raised $3.8 million in pre-seed funding in October last year led by Elkstone.
Zerve provides a visual collaboration platform for data teams that simplifies the building, managing and deployment of data and AI workflows. Teams can set up code blocks that pull in data sources, condition them, control them and feed them through pipelines while working together on a visual canvas without bumping into one another in parallel with automated infrastructure in the background.
"The stick we measure ourselves by is how much more throughput the teams we're working with have been able to achieve, whether that is reducing their overall cycle time for projects or increasing their speed to recommendations," said Zerve Chief Executive Phily Hayes.
The company said infrastructure management, language interoperability, version control, package management and automated deployment processes keep the tedium for data teams low. The platform supports popular languages including SQL, Python and R, and open-source AI and machine learning model datasets such as GPT, BERT and Llama through Hugging Face and AI large language models through Amazon Web Services Inc.'s Bedrock cloud hosted managed service.
Users can also call up Code Assist, an AI integration that helps with code completion, error assistance and debugging. It is connected to the latest LLM models and users can use natural language interactions to generate code suggestions. Users need only comment on what they want done in a block and the assistant will use context awareness of variables, metadata and other code to create a solution. Currently, Code Assist can connect to Anthropic PBC's Claude, OpenAI's GPT models or Mistral AI's models.
Users working with Jupyter Notebooks, a popular web computing platform where users can create and share interactive block-based documents containing code, text and equations with each other to study computational outputs, can import their work directly into Zerve. Programmers, data scientists and students often use Juptyter to document and demonstrate data science and AI coding workflows or perform experiments. It's a common jumping-off point for many projects.
"The speed at which data and AI projects run can sometimes feel like death by 1,000 cuts, so on top of our 100% guaranteed, resilience-based architecture, we have cleared the path-to-output by removing the smaller hurdles that get in the way," said Hayes.
After commercially launching its product in February, the company says that more than 4,000 users have signed up for its free product tier within three months. Although Zerve didn't reveal any customer names, the company said it currently serves several "elite code-first data teams" in major organizations globally.