Construct a knowledge graph: Transform API specifications into a structured graph, then connect previously isolated APIs through business process data.
Intermediate
Knowledge Graphs for AI Agent: API Discovery
Instructors: Pavithra G K, Lars Heling
Earn an accomplishment with PRO
- Intermediate
- Earn an accomplishment with PRO
- Instructors: Pavithra G K, Lars Heling
SAP- Learn more aboutMembership PRO Plan
What you'll learn
Improve API discovery: Do semantic retrieval, and then use business process information from the knowledge graph to discover missing prerequisite APIs and their proper calling sequence.
Build the agent: Create an agent that uses the knowledge graph to discover and execute APIs to carry out real tasks while following the correct process sequence.
Course recap
PROThis course teaches how to use knowledge graphs to help AI agents discover and execute APIs in complex enterprise environments. Taught by Pavithra G K and Lars Heling from SAP Business AI, it addresses the challenge of navigating thousands of APIs with complex interdependencies and business process constraints.
Concept map
Key concepts
- Knowledge graphsGraph-based data models using RDF triples (subject-predicate-object) to represent structured relationships between APIs, entity types, and business processes
- EmbeddingsText embeddings of API entity sets and their properties, indexed with FAISS for semantic similarity search
- RagCombining vector-based API discovery with knowledge graph traversal for business process context
- AgentsLangChain agents equipped with discovery, GET, and POST tools that follow business process sequences
Lesson highlights
- 1.**Introduction** — Problem framing: AI agents need structured API context and process ordering to execute enterprise tasks correctly
- 2.**Knowledge Graphs for AI Agents** — RDF fundamentals, SPARQL queries, knowledge graph construction methods, and enterprise challenges
- 3.**API Knowledge Graph Construction** — Declarative construction from CSV data using SPARQL CONSTRUCT queries, creating 14,000+ triples from API metadata
- 4.**Integration with Business Processes** — Extending the graph with BPMN process data to connect previously disconnected APIs through process activities
- 5.**API Discovery with Knowledge Graphs** — Embedding-based retrieval enhanced with business process graph traversal to find missing dependent APIs
- 6.**Business Process Agent** — Complete LangChain agent with discovery, GET, and POST tools that respects process ordering
- 7.**Conclusion** — Summary and encouragement to apply techniques to own data
About this course
Learn how to help AI agents find and execute the right APIs in the right order using a knowledge graph in “Knowledge Graphs for AI Agent API Discovery”, taught by Pavithra G K (Head of Business Knowledge Graphs) and Lars Heling (Senior Knowledge Engineer) at SAP Business AI.
Large companies may have thousands of APIs, which makes it hard for agents to figure out which APIs to use and in what sequence. This course brings raw API specifications into a knowledge graph and extends it with process data so an agent knows when, and in which order, each API should be called.
You’ll implement API discovery that starts with semantic retrieval to reduce the API selection space, then adds APIs connected by business-process edges. The result is a small, relevant subset of APIs that includes the information about the process sequence the agent should follow.
Finally, you’ll build an agent with a discovery tool, a GET data tool, and a POST data tool. The agent retrieves API metadata (properties, navigations, process information) and executes tasks in the right sequence according to the business process.
In detail, you’ll:
- Understand what knowledge graphs are and how they enable better API discovery and execution in AI applications.
- Construct your first knowledge graph from API services and endpoints and visualize it.
- Extend the knowledge graph with business process data, so APIs are connected via the process and their dependencies.
- Learn how knowledge graphs can be used for API discovery for AI agents; perform semantic retrieval, and then add business-process edges to get all the required APIs and their order.
- Build an agent that uses the knowledge graph to discover and execute APIs following the right process order.
By the end of this course, you’ll have constructed a knowledge graph from API specifications and business process information and built an agent that can take business actions by discovering the right APIs and executing them in the right order.
Who should join?
This course is ideal for developers and AI builders working with large API ecosystems who need their agents to make smarter decisions about which services to call and when. Basic Python knowledge is required.
Course Outline
0 Lessons・0 Code Examples
Elevate your learning experience with Pro
Upgrade to Pro and gain unlimited accomplishments on your resume
Instructors
Course access is free for a limited time during the DeepLearning.AI learning platform beta!
Want to learn more about Generative AI?
Keep learning with updates on curated AI news, courses, and events, as well as Andrew’s thoughts from DeepLearning.AI!