> Markdown version of [/videos/100236-code-is-cheap-software-isn-t?t=1239](https://www.wearedevelopers.com/videos/100236-code-is-cheap-software-isn-t?t=1239). 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).

---

# Code Is Cheap. Software Isn’t.

Generating syntax with AI is cheap. Building resilient enterprise software is not. Learn how to tame coding agents with strict guardrails and dynamic context.

- **Speakers:** [Markus Eisele](https://www.wearedevelopers.com/@markus-eisele)
- **Event:** World Congress 2026 Europe
- **Published:** July 10, 2026
- **Duration:** 29:55
- **URL:** https://www.wearedevelopers.com/videos/100236-code-is-cheap-software-isn-t

## Summary

The era of AI coding agents has revealed a stark reality: code generation is cheap, but resilient software remains incredibly expensive. While agents can effortlessly spin up boilerplate, they consistently stumble in large, legacy enterprise systems where implicit knowledge, hidden constraints, and complex cross-module configurations dictate success. Without explicit context, ambiguity inevitably turns into hallucinations. The core challenge of software engineering has simply shifted from writing syntax to rigorously defining system intent, constraints, and architecture. To integrate AI safely into long-lived codebases, developers must stop relying on clever one-off prompts and instead construct robust, repeatable workflows that clearly separate what needs to be achieved from how it should be implemented.

Treating AI-generated code as a pull request from an untrusted junior teammate is crucial for maintaining control over enterprise systems. Success requires enforcing strict guardrails through static validation tools like PMD, pre-commit hooks, and explicit architectural documentation such as Architecture Decision Records (ADRs). Context management is equally important; instead of overwhelming the agent's context window with entire wikis, tools like Grounded Docs, Context 7, and the Quarkus Agent MCP can dynamically inject highly focused, runtime-specific knowledge into the workflow. By structuring interactions via discrete skills and commands, teams can transform fragile prompt engineering into disciplined agentic software development.

Ultimately, traditional decomposition and agile principles are more vital than ever when delegating tasks to AI. The "one Monday morning rule" dictates that if an assignment cannot be explained to the agent in a single sentence, it is simply too large. Developers must return to small, iterative changes anchored by tests—where tests serve as the ultimate guardrails for autonomous agents. As cognitive overload becomes a dominant byproduct of accelerated code generation, engineering teams must prioritize rigorous verification over raw speed, ensuring that the relentless pace of AI does not compromise software reliability or human mental health.

**Keywords:** AI coding agents, legacy java modernization, architecture decision records, static code validation, pre-commit agent hooks, Quarkus agent MCP, IBM bob assistant, grounded docs MCP, context window optimization, code generation hallucinations, developer cognitive overload, agentic software development, prompt engineering constraints, test-driven agent workflows, enterprise software maintenance

## Chapters

1. **The reality of using AI coding agents in legacy codebases** (01:10) — Coding agents struggle to infer hidden constraints and invisible knowledge scattered throughout large enterprise applications.
1. **How missing context leads to subtle AI agent hallucinations** (04:36) — AI models invent plausible but incorrect solutions when they lack full visibility into complex enterprise architectures.
1. **Managing architecture ambiguity using explicit coding agent instructions** (07:37) — Providing explicit architectural workflows and clear documentation prevents agents from making hazardous structural assumptions.
1. **Delegating tasks by separating intent from final agent implementation** (10:26) — Separating the definition of intent from the implementation details enables static verification of generated results.
1. **Defining strict task boundaries and using tests as safety guardrails** (13:09) — Starting with tests and strict dependency limits creates firm boundaries that restrict overly creative behavior.
1. **Decomposing large agent requests into small agile software iterations** (15:01) — Breaking complex transformations into small single-step iterations keeps AI code outputs transparent and manageable.
1. **Using architecture decision records to formalize agent operating rules** (17:26) — Documenting the reasoning behind technical choices guides coding agents effectively throughout entirely long-lived software projects.
1. **Turning repetitive developer tasks into automated AI agent skills** (20:39) — Structuring repetitive review and refactoring processes into predictable agent skills avoids the pitfalls of endless conversations.
1. **Enhancing agent accuracy with model context protocol servers** (21:59) — Connecting agents to documentation indexes and live runtime endpoints significantly improves the relevance of their suggestions.
1. **Structuring AI-driven product development into explicit phased implementation workflows** (24:49) — Executing AI modifications in discrete phases of exploration, decision, and review maximizes control over systematic implementation.
1. **Treating generated code as untrusted contributions to prevent developer burnout** (26:15) — Strictly reviewing AI-generated pull requests maintains system integrity while concurrently managing the cognitive fatigue of constant supervision.

## Related Moments

- [Balancing developer autonomy with the adoption of coding agents](https://www.wearedevelopers.com/videos/100198-the-last-mile-of-ai-from-prototype-to-production) (from "The Last Mile of AI: From Prototype to Production")
- [Motivations for adopting AI to enhance developer productivity](https://www.wearedevelopers.com/videos/1266-navigating-the-ai-revolution-in-software-development) (from "Navigating the AI Revolution in Software Development")
- [The promise and risk of AI coding agents](https://www.wearedevelopers.com/videos/100277-what-production-knows-closing-the-loop-between-ai-agents-and-the-systems-they-build) (from "What Production Knows: Closing the Loop Between AI Agents and the Systems They Build")
- [Shifting developer workloads and realistic AI productivity gains](https://www.wearedevelopers.com/videos/1830-wearedevelopers-live-speculaitions) (from "WeAreDevelopers LIVE - SpeculAItions")
- [Balancing AI tool mandates with developer trust and productivity](https://www.wearedevelopers.com/videos/1365-wearedevelopers-live-the-weekly-developer-show-with-chris-heilmann-and-daniel-cranney) (from " WeAreDevelopers LIVE - the weekly developer show with Chris Heilmann and Daniel Cranney")
- [Rethinking team structures around AI agent capabilities](https://www.wearedevelopers.com/videos/1539-agentic-devops-how-ai-powered-automation-transforms-software-delivery-on-github-and-azure) (from "Agentic DevOps: How AI-Powered Automation Transforms Software Delivery on GitHub and Azure")

## Related Articles

- [What is Agentic Programming and Why Should Developers Care?](https://www.wearedevelopers.com/magazine/625-what-is-agentic-programming-and-why-should-developers-care)
- [Never delegate the understanding](https://www.wearedevelopers.com/magazine/749-never-delegate-the-understanding)
- [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)
- [One billion (bad?) developers: How AI is changing the way we learn to code](https://www.wearedevelopers.com/magazine/516-one-billion-bad-developers-how-ai-is-changing-the-way-we-learn-to-code)

## Related Jobs

- [Principal Product Manager, Agent Platform](https://www.wearedevelopers.com/jobs/ext/277541-principal-product-manager-agent-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**
- [Principal Software Engineer, Enterprise AI Platform](https://www.wearedevelopers.com/jobs/ext/1467292-principal-software-engineer-enterprise-ai-platform) at **GitHub**
- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [Senior Software Engineer, Enterprise Products](https://www.wearedevelopers.com/jobs/ext/1841248-senior-software-engineer-enterprise-products) at **GitHub**