​
• Understanding the Core Connection

The Chronicle Data is one of the central and inevitable ledgers within Ekyam. These agents are responsible for recording (reading) new events and retrieving past data events (writing). How do these Agents interact with Core Chronicle Data AI Agents interact with the Core Chronicle Data via a secure, two-part system: Writing new data and Reading it as well. This interaction is governed by a read-only model via MCP server and the Knowledge Graph is exposed via the tool to it. → Reading from Chronicle: Agents use the knowledge graph for performing complex queries. Since the Knowledge Graph is an interconnected representation of the data in the chronicle, the agent can query the graph for relationships instead of searching several log events.  A key component of the retrieval process, The Knowledge Graph, is hosted on Neo4j on Google Cloud, which requires valid credentials and security protocols for access.

​
• Querying Retail Knowledge Graph

A Retail Knowledge Graph is a repository that connects the key retail entities like products, customers, stores and suppliers in a semantic network. The agents interact with the RKG to follow a multi-hop reasoning with a chain of relationships to gather complex and rich information. This showcases the role of an AI agent from a search tool to an intelligent response engine.  Demonstrating Agent-Graph Interaction Agents query the RKG using Cypher queries that traverse multiple relationships types ((eg. MATCH (c:Customer)[:PURCHASED]> (p:Product) [:BELONGS_TO] (cat: Category)). This enables them to retrieve contextual information like customer preferences, product associations, and purchase patterns in a single query rather than multiple database calls.  Examples of Complex Queries Take a look at below mentioned examples of complex queries that leverage the graph to answer high-level business questions: →Customer segmentation: MATCH  (c:Customer)[PURCHASED]>(p: Product)[:BELONGS_TO]>(cat: Category) WHERE cat.name IN [‘Premium’, ‘Luxury’] RETURN c.demographic, COUNT(DISTINCT p) as purchase_count ORDER BY purchase_count DESC - identifies high-value customers by their premium product purchases. → Cross-Selling Opportunities: MATCH (c:Customer)[:PURCHASED]>(p1:Product)[:FREQUENTLY_BOUGHT_WITH]>(p2:Product) WHERE NOT(c)[:PURCHASED]>(p2) RETURN c.id, p1.name, p2.name - finds customers who bought one product but not its frequently co-purchased companion. → Seasonal Trend Analysis: MATCH (p:Product)[:SOLD_IN]> (t:Transaction):OCCURRED_IN WHERE s.name= ‘Holiday’ RETURN p.category, AVG(t.quantity) as avg_quantity, SUM(t.revenue) as total_revenue - analyzes holiday season performance across product categories.