
Getty Images/iStockphoto
Knime updates framework for agentic AI development
The open source analytics vendor is keeping up with competitors by providing capabilities aimed at enabling users to create cutting-edge applications capable of autonomous action.
Knime is in on agentic AI development.
Like peers such as Domo and ThoughtSpot, Knime first unveiled a framework for building agentic AI tools during its Spring Summit in March. The vendor’s latest platform update adds new capabilities to the framework that make it easier for users to build agents.
Agents are the dominant trend in AI. Unlike generative AI-powered chatbots that are reactive, agents are AI applications that have contextual awareness and can reason on their own. As a result, agents can act autonomously to assist workers.
Given that the update -- which includes no-code capabilities -- makes it easier for users to build agents, Knime's updated framework for agentic AI development is important for users, according to Mike Leone, an analyst at Enterprise Strategy Group, now part of Omdia.
"It's a significant leap forward for Knime customers, helping them simplify and accelerate AI agent development while making it accessible to a broader range of users," he said. "They can now create agent tools directly from existing workflows with no-code ease."
Based in Zurich, Switzerland, with a U.S. headquarters in Austin, Texas, and offices in Germany, Knime is an analytics vendor whose open source platform includes data science capabilities. In July 2024, Knime's platform update focused on AI governance.
New capabilities
Three years after OpenAI's November 2022 launch of ChatGPT, which represented a significant improvement in generative AI technology, enterprises continue to increase their investments in developing AI tools, given their potential to make workers better informed and more efficient.
Initially, chatbots that enable users of all technical skill levels to query and analyze were the vanguard following ChatGPT's launch. AI, however, has evolved since then, and now, after emerging in 2024, agents represent the forefront.
Given the vital role data plays in AI development, many data management and analytics vendors have created AI development environments to make it easier for customers to securely combine proprietary data with AI models. In recent months, with agents becoming the dominant trend in AI, they've added capabilities specifically designed to simplify agentic AI development.
In addition to analytics specialists such as Domo, Knime and ThoughtSpot, data platform vendors Databricks and Snowflake have each added features specific to agentic AI development, as have tech giants such as AWS and Google Cloud.
After making its initial foray into enabling agentic AI development in March, Knime is adding depth to its development framework.
Rather than forcing users to learn a new workflow, Knime's update, released earlier this month, lets developers create agents using the same concepts they use to develop other applications. The update also adds no-code tools that enable new users to build agents easily.
Among the new capabilities are Agent Prompter to select and call tools such as models and data files during development, an Agent Chat View node for users to chat with agents and ask them to perform certain tasks, a simple metadata structure to aid governance and free courses to support users as they build and deploy agents.
With its updated agentic AI development framework, Knime is providing a valuable set of capabilities, according to Kevin Petrie, an analyst at BARC U.S.
"Knime is definitely strengthening its competitive position in the market," he said. "They have enriched how they help citizen data scientists and developers build AI applications and agents, with a focus on extensibility, model choice and open ecosystem support. They have a strong, longstanding user community that has grown over time, and I expect their users will value these new capabilities."
Of particular benefit is the way Knime now organizes metadata for conversations and interactions between users, models and agents, which aids governance as agentic workflows grow more complex, Petrie continued.
While Knime's update aims to make it easier for users to build agents, customer feedback provided only part of the impetus for updating its agentic AI development framework, according to Michael Berthold, the vendor's co-founder and CEO.
"It was a combination of user feedback and forward-thinking design," he said. "We listened closely to our community while also exploring how to simplify agentic AI development."
In addition to the agentic AI development framework, Knime's update includes the following:
- Access to AI models from Anthropic, Google and IBM in addition to those from OpenAI, Microsoft and Hugging Face that customers could previously use when developing AI tools.
- User interface improvements including a side panel that enables users to configure nodes -- connectors for writing, reading and moving workflows -- without toggling between environments.
- Direct connectivity to Microsoft Fabric that enables users to manage their Power BI semantic models as well as read and write data within Microsoft Fabric OneLake and Data Warehouse without leaving their Knime environment.
While overshadowed by the update agentic AI development framework, Knime's user interface improvements and the integration with Microsoft Fabric are significant, according to Leone.
"They stand to enhance the everyday interaction for all users," he said.
Looking ahead
With its latest update now generally available, Knime's product development focus remains on AI -- including agents -- and users' workflow experience, according to Berthold.
Regarding AI and agents, the vendor aims to continue making it easier to develop advanced applications, integrate AI into users' workflows to improve their productivity and add support for multimodal AI, he said.
Continuing to focus on AI and agents is wise, according to Leone, who noted that as agentic AI evolves, it will be important for vendors to provide more than a basic framework for developing agents.
"Moving AI agent capabilities and governance beyond the basics is important," he said. "The current framework is excellent for individual agents, but the future lies in multi-agent systems where agents can collaborate, delegate tasks and even learn from each other.
Adding support for those more complex agentic behaviors will therefore be important, Leone continued.
Petrie, meanwhile, noted that while Knime provides strong capabilities for integrating and operating the data, model and application lifecycles, it could do more to help customers govern their AI initiatives.
"I'd be interested to see them build more partnerships with catalog and data quality vendors to assist governance, which is fast becoming the biggest challenge of agentic AI initiatives," he said.
Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than 25 years of experience. He covers analytics and data management.