InfraNodus + Crew AI: Enhancing AI Agent Workflows with Content Gap Detection & Research Questions

You can use InfraNodus API to augment your AI agent workflows with the knowledge graph insights.  InfraNodus can be especially useful for content gap detection and research question generation.

A typical use case would be the need to study a certain topic and to understand what's missing in the current discourse and how this topic could be developed further. It is of course possible to create a standard LLM agent that will be tasked with identifying a gap in the discourse, but the results obtained will be quite generic.

InfraNodus API can help avoid the LLM's bias and use the underlying graph structure to identify the topical clusters that are not yet connected in the existing discourse. These clusters can then be used to generate a content gap or research questions that can be used by AI agents to develop the discourse further. 

You can find the full source code you can use for your CrewAI agent workflow in our GitHub repo: https://github.com/infranodus/crewai-infranodus-templates

Crew #1: Generate a Content Gap based on the Underlying Knowledge Graph 

In this setup, we augment the original research demo of Crew AI with an additional InfraNodus agent that uses the `infranodus_gap_finder` tool to identify the gap that exists in a certain topic.

1. The first agent performs a research on the topic you specify

2. The second agent will send the results of the research to the InfraNodus API (via the dotGraphFromText endpoint) to retrieve the clusters with the highest gap between them in the underlying knowledge graph.

3. The gap will then be used by the second agent to distill a description of the gap in the topic you're studying

4. The description of the gap will then be used to generate a report on how this gap in the topic can be addressed.

Crew #2: Generate a Question Based on the Knowledge Graph Gap

In this setup, we augment the original research demo of Crew AI with an additional InfraNodus agent that uses the `infranodus_question_generator` tool to generate a research question.

1. The first agent performs a research on the topic you specify

2. The second agent will then generate a research question using the built-in InfraNodus tool. It will extract the research questions from the InfraNodus API (via the graphAndAdvice endpoint) as well as the underlying graph structure and structural gaps.

3. It will then use LLM to distill this information into a single question

4. The question will be passed to the research agent who will then use LLM to generate a report on how this question can help develop the field you study.

Create an account on https://infranodus.com to try these workflows now.

The source code to use for your CrewAI agents can be found in our GitHub repo: https://github.com/infranodus/crewai-infranodus-templates