o9’s composite agents are combined with LLM systems trained on business-specific information.
Enterprise AI software provider o9 has announced an upgrade to its Digital Brain platform with the integration of genAI-powered Large Language Model (LLM) composite agents. These advanced agents are poised to transform the way planners handle complex tasks, further advancing the capabilities of integrated business planning.
“Our Digital Brain’s EKG offers cross-functionality as it touches supply chain, finance, procurement, commercial, customers and suppliers. It connects the knowledge of all these entities – and is constantly accumulating new knowledge as well. Now we are deploying composite agents that can do a lot more cross-functional analysis,” said Anand Srinivasan, Chief Strategy Officer, o9.
o9’s composite agents are constructed on a base of atomic agents—AI-driven systems adept at executing tasks, retrieving information, and generating responses based on specific inputs. These are combined with LLM systems trained on business-specific information.
Also Read: Salesforce Unveils LLM Benchmark to Optimise GenAI Implementations
Within the core Enterprise Knowledge Graph (EKG) of the o9 Digital Brain platform, composite agents can sequence atomic agents to perform intricate tasks necessary for cross-functional planning processes.
Examples include developing forecasts or conducting post-game analyses by synthesising data to derive insights, such as comparing month-to-month forecast variations or analysing past quarter performance against forecasts to identify root causes of discrepancies. In addition, these agents are trained on the ‘recipes’ used by an organisation’s planning experts and continuously improve through feedback, enhancing their ability to achieve desired outcomes.
“Typically, an employee within a particular function only has the purview of knowledge, people, and systems within their department; they don’t look across and consider, ‘How does my decision impact someone else in a different department or organisation?’ Hence, a lot of things get missed, and there is a lot of value leakage. This is another thing these composite agents can help solve by becoming more cross-functional in their analysis,” added Srinivasan.