> Markdown version of [/videos/1417-new-ai-centric-sdlc-rethinking-software-development-with-knowledge-graphs?t=715](https://www.wearedevelopers.com/videos/1417-new-ai-centric-sdlc-rethinking-software-development-with-knowledge-graphs?t=715). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# New AI-Centric SDLC: Rethinking Software Development with Knowledge Graphs

After massive context windows failed to refactor legacy code, one engineering team pivoted to GraphRAG. See how knowledge graphs cut through architectural complexity to build an AI-centric SDLC.

- **Speakers:** [Gregor Schumacher](https://www.wearedevelopers.com/@gregor-schumacher), [Marcel Gocke](https://www.wearedevelopers.com/@marcel-gocke), [Sujay Joshy](https://www.wearedevelopers.com/@sujay-joshy)
- **Event:** World Congress 2025
- **Published:** August 20, 2025
- **Duration:** 15:26
- **URL:** https://www.wearedevelopers.com/videos/1417-new-ai-centric-sdlc-rethinking-software-development-with-knowledge-graphs

## Summary

When tasked with increasing software engineering productivity by 15 percent, the internal tech team realized that leaning on standard AI coding tools would not suffice. Their analysis revealed that writing actual code comprised only about 13 percent of the software engineering process, while the vast majority of time was spent retrieving information, handing over tasks, and digging through corporate context. Initial experiments attempting to refactor entire applications by dumping messy code into models with massive context windows ultimately failed. The large language models either hallucinated or struggled to find the necessary needles in the corporate haystack, underscoring that simply expanding context windows is not a viable silver bullet for complex legacy transformations.\n\nThe failure of massive context stuffing prompted a pivot away from flat document stores and standard vector databases. Vector embeddings proved ineffective for large codebases because similar branches or copy-pasted files generated nearly identical embeddings, confusing the language model's statistical retrieval. Instead, the team revived abstract syntax tree concepts to build a graph-based retrieval augmented generation (GraphRAG) architecture. By mapping classes, methods, database calls, and vast corporate documentation into a cohesive knowledge graph, the system can pinpoint exact endpoints and dependencies. This targeted approach isolates the specific call chains needed, feeding concise, highly relevant context to the AI rather than overwhelming it.\n\nExpanding this concept across the entire tech stack transformed how developers interact with over 600 internal tools, moving the AI integration beyond code and into architecture and product specifications. To manage the complexity of reasoning models and multi-agent systems, the platform introduced a knowledge graph of thought to audit and orchestrate autonomous tasks. By tracking every action an agent takes—whether adjusting a Jira ticket or querying Confluence—the orchestration layer pumps precise context to the right agents at the right time. This multi-layered visibility ensures reproducibility, reduces hallucination across multiple hops of complexity, and secures a scalable, AI-centric software development lifecycle.

**Keywords:** software development lifecycle, knowledge graph abstraction, graphrag architecture, abstract syntax tree maps, enterprise code refactoring, multi-agent orchestration, knowledge graph of thought, context window limitations, vector database challenges, legacy system transformation, developer productivity metrics, corporate context retrieval, ai hallucination reduction, information retrieval optimization, software dependency mapping

## Chapters

1. **Focusing on process disruption for developer productivity** (00:24) — Analyzing software engineering processes reveals that discovering information consumes more time than writing code.
1. **Evaluating large context windows for code refactoring** (01:50) — Dumping massive codebases into large language models fails due to complicated dependencies and missing corporate context.
1. **Replacing flat vector databases with abstract syntax trees** (04:41) — Storing code structures in a graph rather than flat vector databases improves accuracy by capturing exact relationships.
1. **Building a centralized knowledge graph for corporate codebases** (08:47) — A centralized knowledge graph parses interconnections to pinpoint specific code dependencies across hundreds of legacy repositories.
1. **Orchestrating AI agents with a knowledge graph of thought** (11:55) — An orchestration platform feeds relevant context to specialized agents while recording all actions into an auditable graph.

## Related Moments

- [Shifting developer workloads and realistic AI productivity gains](https://www.wearedevelopers.com/videos/1830-wearedevelopers-live-speculaitions) (from "WeAreDevelopers LIVE - SpeculAItions")
- [Introduction to generative AI and knowledge graphs](https://www.wearedevelopers.com/videos/1154-large-language-models-knowledge-graphs) (from "Large Language Models ❤️ Knowledge Graphs")
- [The growing necessity of orchestrating AI in software teams](https://www.wearedevelopers.com/videos/1950-the-scrum-master-as-an-orchestrator-guiding-human-ai-collaboration-in-modern-teams) (from "The Scrum Master as an Orchestrator: Guiding Human–AI Collaboration in Modern Teams")
- [Applying context engineering across the full software lifecycle](https://www.wearedevelopers.com/videos/100256-can-this-elephant-dance-ibm-bob-and-the-future-of-ai-first-software-development) (from "Can This Elephant Dance? IBM Bob and the Future of AI-First Software Development")
- [Discussion on AI hallucinations and practical developer workflows](https://www.wearedevelopers.com/videos/805-openai-for-fintech-building-a-stock-market-advisor-chatbot) (from "OpenAI for FinTech: Building a Stock Market Advisor Chatbot")
- [Exploring AI integrations in modern agile development workflows](https://www.wearedevelopers.com/videos/631-chatgpt-create-a-presentation) (from "ChatGPT: Create a Presentation!")

## Related Articles

- [Graph and AI Trends 2026: Why Is AI Running but Not Yet Delivering?](https://www.wearedevelopers.com/magazine/680-graph-and-ai-trends-2026-why-is-ai-running-but-not-yet-delivering)
- [Transforming Software Development: The Role of AI and Developer Tools](https://www.wearedevelopers.com/magazine/527-transforming-software-development-the-role-of-ai-and-developer-tools)
- [What is Agentic Programming and Why Should Developers Care?](https://www.wearedevelopers.com/magazine/625-what-is-agentic-programming-and-why-should-developers-care)
- [How we Build The Software of Tomorrow](https://www.wearedevelopers.com/magazine/120-how-we-build-the-software-of-tomorrow)

## Related Jobs

- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [Principal Product Manager, Agent Platform](https://www.wearedevelopers.com/jobs/ext/277541-principal-product-manager-agent-platform) at **GitHub**
- [Principal Software Engineer, Enterprise AI Platform](https://www.wearedevelopers.com/jobs/ext/1467292-principal-software-engineer-enterprise-ai-platform) at **GitHub**
- [Senior AI Agent Software Engineer (Go, Python) (m/f/x)](https://www.wearedevelopers.com/jobs/48277-senior-ai-agent-software-engineer-go-python-m-f-x) at **Dynatrace**
- [Staff Software Engineer, Copilot Experiences](https://www.wearedevelopers.com/jobs/ext/164361-staff-software-engineer-copilot-experiences) at **GitHub**
- [Head of AI Applications](https://www.wearedevelopers.com/jobs/ext/1456210-head-of-ai-applications) at **ZEISS Group**