Domo unveils agentic AI toolkit to simplify development
Agent Catalyst provides users with a four-step process for developing agentic applications, including selecting an LLM and connecting it to relevant proprietary data.
Agent Catalyst, a new toolkit from Domo designed to simplify agentic AI development, is now generally available.
Unlike AI assistants and chatbots that respond to queries, agents are generative AI applications able to act autonomously. Among their capabilities are surfacing insights and recommending actions, as well as performing certain repetitive tasks such as documentation and code generation.
Agents have become the dominant trend in AI development over the past year because they can make both nontechnical and trained data experts better informed and more efficient.
Agent Catalyst was introduced on March 19 during Domopalooza -- Domo's annual user conference in Salt Lake City -- and is now generally available. The toolkit is a suite within Domo.AI, the vendor's environment for generative AI development capabilities.
Agent Catalyst enables Domo customers to build generative AI agents by selecting a large language model (LLM) and connecting it to the proprietary enterprise data appropriate for the application.
Given that Agent Catalyst makes agentic AI development a straightforward process, it's an important addition for Domo users, according to Michael Ni, an analyst at Constellation Research.
Catalyst is a game changer, providing a structured framework for building and deploying AI agents without deep coding expertise.
Michael NiAnalyst, Constellation Research
"Catalyst is a game changer, providing a structured framework for building and deploying AI agents without deep coding expertise," he said. "It enables users to leverage Domo's data and workflow infrastructure to create custom AI solutions tailored to business needs, lowering the barrier to AI adoption."
Based in Silicon Slopes, Utah, Domo is a cloud-based analytics vendor that, like peers including Qlik and MicroStrategy, has expanded into generative AI development over the past two years.
Building agents
In the first year of the ongoing explosion of interest in generative AI development that began with OpenAI's November 2022 launch of ChatGPT, assistants and chatbots represented the cutting edge.
By combining LLMs with proprietary data, enterprises were able to develop applications that enable users to query and analyze data using free-form natural language rather than business-specific language or code. By doing so, they were able to expand data analysis beyond experts to virtually any employee who could benefit from data, making their workers better informed.
Given data's role in AI development, many data management and analytics vendors, including Domo, responded to rising interest in AI by providing tools that simplify developing assistants and chatbots.
By early 2024, agentic AI emerged. Now, just as data management and analytics vendors enable developing assistants and chatbots, many are providing tools to develop agents. For example, Tableau is redesigning its platform around agentic AI, while Databricks and Snowflake are creating AI development platforms in addition to their data management capabilities.
Agent Catalyst, working in concert with previously existing Domo.AI features, is Domo's version of an agentic AI development suite and represents the progression of Domo's foray into AI development, according to Ben Schein, the vendor's senior vice president of product.
Regarding the impetus for the suite's development, Domo noticed that some customers were using components within Domo.AI to develop agents, leading to Domo putting them together in one environment, he continued.
"Domo wanted to provide an easier way to quickly create agents rather than the more complicated steps that were required from early AI adopters," Schein said. "Agent Catalyst provides the tooling to more simply create agents."
Agent Catalyst enables Domo users to develop generative AI agents in four steps: selecting the LLM that fits their needs, providing instructions to the LLM, connecting the LLM to data and enabling the agent to access appropriate tools.
New features designed to assist during the four steps include the following:
DomoGPT, Domo's generative AI language model, which can be used in conjunction with other language models.
FileSets, a tool for managing unstructured data such as text and audio that feeds retrieval-augmented generation pipelines.
Semantic layer features that enable users to define data to ensure consistency and make it discoverable.
AI assistant and agent builders that further simplify development.
DomoGPT and FileSets are now available, while the semantic layer features and AI development builders are expected to be generally available later this year.
Although some Agent Catalyst capabilities are still under development, Domo nevertheless provides a competitive platform that combines analytics and AI development capabilities, according to Matt Aslett, an analyst at ISG Software Research. Among them are tools that enable analysis through natural language queries, guided analysis, assisted data preparation and integration, and AI model management.
"Domo offers capabilities that transcend traditionally separate product categories to address wider analytics and data challenges, including data integration, workflow and application development," Aslett said.
Meanwhile, once available, the semantic layer will be "particularly beneficial," according to Ni, because it will enable business users to contextualize data, improving the accuracy and relevancy of AI outputs.
"This feature is crucial for refining chat experiences and ensuring that AI agents understand and operate within the specific business context," he said.
Likewise, Aslett highlighted the semantic layer as an important feature of agentic AI development.
"Semantic layer capabilities will enable enterprises to ensure agentic AI applications incorporate standardized business definitions and metrics," he said.
In addition to Agent Catalyst, Domo unveiled other new AI capabilities during Domopalooza, including automated insight generation, AI Agent Tasks to execute processes with agents, and generative AI-powered data transformation and enrichment.
Next steps
With Agent Catalyst generally available, ease of use will be a focal point of Domo's roadmap, according to Schein.
Ease-of-use initiatives include adding a navigation feature now in beta testing that learns from recent user activity to personalize experiences by surfacing relevant content.
More AI technology is also part of Domo's roadmap, including adding features to its semantic layer and the addition of more agents within Domo's platform to help users with tasks such as building dataflows and applications, Schein added.
Ni, meanwhile, noted that Domo has been wise in expanding the reach of its platform through tools such as Cloud Amplifier, which enables integrations with cloud data storage platforms, and should continue working to strengthen partnerships with cloud data platforms.
"User experience wins will be the attention grabbers for customers," 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.
