(Tech Talk) Shipping with Context Knowledge Graphs as the Backbone of AI-First Software Delivery

Knowledge graphs are essential to solving the context bottleneck in AI-First software delivery, which occurs because workflows, policies, and dependencies are siloed and invisible to AI agents.

In this Tech Talk, Prateek Mittal ((Product Director of AI Core and Data Platform at Harness)) discusses the key concepts:
Knowledge Graphs vs. Observability: Observability tells you "what is happening," while knowledge graphs tell you "what does that mean" by modeling structured relationships. They work together to link live signals to affected services or SLAs.

AI Operational DevOps: This approach uses AI to safely move code from development to production, relying on a platform approach powered by the knowledge graph.4

Preventing Failure: Avoid overmodeling, undermodeling, and using stale data. The data must be near real-time.

Guardrails: Security is maintained by applying policies (e.g., PII protection) and using Role-Based Access Control (RBAC), treating the AI agent as an extension of the human with the principle of least privilege.

Starting Small: Begin by ingesting Git, CI/CD, and telemetry data. The smallest viable use case is automated Root Cause Analysis (RCA).

The knowledge graph is a shared model owned by Platform Engineering, SRE, Security, and App teams, ensuring holistic context. Harness uses its Software Delivery Knowledge Graph to power its unified agent across FinOps, DevOps, and Security Ops