KNIME, the open source data analytics and AI company, is releasing its AI companion, K-AI, enabling users to co-create powerful data workflows with AI.According to the company, K-AI will answer questions, make recommendations, and extend or build whole data workflows based on user prompts. The AI companion speeds up the time to insight, while giving users complete transparency into and control over what AI is doing.
This is a major step forward in the automation of data work that helps users of any skill level derive better insights from their data, faster, the company said.
KNIME’s intuitive visual workflows help data workers immediately understand how AI contributed to their work. The visual workflow acts as documentation of every step in a data manipulation or data analysis process, as well as also adding its own narrative text to document its work, so users can easily review it.
The vendor said, the workflow itself represents a paradigm shift for data workers using AI, from opaque code-heavy environments to transparent documented workflows anyone can understand.
K-AI aims to assist anyone in the organization upskill faster to do data work, taking the burden off internal experts.
K-AI offers two modes:
- Q&A Mode: Users can ask KNIME-related technical or theoretical questions, and K-AI provides answers, helping users better understand how specific nodes work or how to further improve their workflows.
- Build Mode: K-AI collaborates with users in the canvas by adding, connecting, and configuring nodes based on user interactions. Users can build a simple workflow with a single prompt or get K-AI’s help to advise on the next node during workflow creation. K-AI can also help in styling visualizations, generating embedded Python or R scripts, and more. K-AI always explains its work so humans can see what has been added to their workflows and why.
In build mode, K-AI can extend existing workflows and in Q&A mode it can recommend extensions, workflows, and nodes from KNIME Community Hub. This makes it easier than ever for users to discover existing solutions and build on community contributions without starting from scratch, according to the company.
"KNIME’s mission has always been to enable anyone to be able to make sense of their data, regardless of their technical background," said Michael Berthold, CEO of KNIME. "With K-AI, we’re taking a significant step forward in helping individuals build workflows faster and learn more in the process. By combining AI assistance with KNIME’s open-source platform, we’re not only accelerating how users work with data but also empowering them to learn and grow as they go. We’re excited to see how K-AI will help users unlock new opportunities with their data."
K-AI was first released as an experimental extension in early 2024. During the experimental phase, both the K-AI Q&A mode and the Build mode were tested and refined. Now K-AI is for the first time released into KNIME for all users as part of the broader KNIME Analytics Platform 5.4 release, which includes numerous improvements to LLM connectivity, LLM evaluation capabilities, and the overall user experience:
- Improved AI extension with support for Databricks LLMs and GPT4All models on IBM Granite and Microsoft Phi 3.5 on top of existing support for OpenAI, Azure, Hugging Face, and Ollama models.
- The ability to detect vulnerabilities such as bias, hallucination, data leakage and more in AI-augmented workflows with Giskard LLM and Giskard RAG nodes.
- Multiple connectivity improvements to Tableau, Box, and Databricks on top of KNIME’s 300-plus existing data source integrations.
Starting with KNIME Analytics Platform 5.4, K-AI is included for all users.
Users can also try building with K-AI directly in the browser via KNIME’s new playground environment.
For more information about these updates, visit www.knime.com.