Join our daily and weekly newsletters for the latest industry-leading AI updates and exclusive content. learn more
Popular AI orchestration frameworks llama index goes beyond the Search Augmentation Generation (RAG) process and introduces Agent Document Workflow (ADW), a new architecture that improves agent productivity.
As orchestration frameworks continue to improve, this approach could provide organizations with the option to enhance the decision-making capabilities of their agents.
According to LlamaIndex, ADW helps agents manage “complex workflows that go beyond simple extraction and matching.”
Some agent frameworks are based on RAG systems, which provide agents with the information they need to complete their tasks. However, this method does not allow the agent to make decisions based on this information.
LlamaIndex showed some real-life examples of how ADW works well. For example, contract reviews require human analysts to extract critical information, cross-reference regulatory requirements, identify potential risks, and generate recommendations. When deployed in that workflow, an AI agent would ideally follow the same pattern and make decisions based on the documents it reads for contract review and knowledge from other documents.
“ADW addresses these challenges by treating documents as part of a broader business process,” LlamaIndex said in the article. blog post. “ADW systems can not only analyze document content, but also maintain state across multiple steps, apply business rules, coordinate various components, and take actions based on document content.”
LlamaIndex previously said that while RAG is an important technology, it remains primitive, especially for companies seeking more robust decision-making capabilities using AI.
Understand the context of decisions
LlamaIndex has developed a reference architecture that combines LlamaCloud analytics and agents. “Build systems that can understand context, maintain state, and drive multi-step processes.”
To do this, each workflow contains a document that acts as an orchestrator. Tap LlamaParse to extract information from your data, maintain document context and process state, and direct agents to retrieve reference material from another knowledge base. From here, agents can begin generating recommendations for contract review use cases and other actionable decisions for various use cases.
“Maintaining state throughout the process enables agents to handle complex multi-step workflows that go beyond simple extraction and matching,” the company said. “This approach allows us to build deep context about the documents we are processing while coordinating between the various system components.”
Different agent frameworks
Agent orchestration is an emerging field, and many organizations are still exploring how an agent (or agents) works. AI agent and application orchestration is likely to become a bigger topic this year as agents move from single systems to multi-agent ecosystems.
The AI agent is an extension of the capabilities provided by RAG: the ability to search for information based on enterprise knowledge.
But as more companies start deploying AI agents, they also want them to perform many of the tasks that human employees do. And for these more complex use cases, a “standard” RAG is not enough. One advanced approach that companies are considering is agent RAGs, which extend the agent’s knowledge base. Before getting a result, the model can decide whether more information needs to be found, what tools to use to get that information, and whether the context it just got is relevant.
