> Markdown version of [/videos/1941-25-years-after-agile-and-yet-we-re-building-waterfall-agents-martin-hynie?t=52](https://www.wearedevelopers.com/videos/1941-25-years-after-agile-and-yet-we-re-building-waterfall-agents-martin-hynie?t=52). 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).

---

# 25 Years After Agile, And Yet We're Building Waterfall Agents - Martin Hynie

Martin Hynie warns that modern AI agents have regressed into rigid waterfall workflows. They blindly pursue flawed logic. Discover how ritualized dissent forces these autonomous systems to self-correct.

- **Speakers:** Martin Hynie
- **Event:** Coffee With Developers
- **Published:** June 1, 2026
- **Duration:** 36:56
- **URL:** https://www.wearedevelopers.com/videos/1941-25-years-after-agile-and-yet-we-re-building-waterfall-agents-martin-hynie

## Summary

Although Agile methodologies promised adaptable software development, modern AI agents executing long-running tasks have regressed into a rigid waterfall approach. When given a complex prompt, current agentic workflows typically form an initial strategy and relentlessly pursue it, making minor iterative adjustments rather than fundamentally rethinking the problem. This lack of strategic flexibility prevents agents from identifying when they are on the wrong path, a flaw amplified by the "sycophantic" nature of LLMs that confidently pursue flawed logic rather than exhibiting doubt or humility.

To introduce complexity theory and self-correction into autonomous systems, the Strategic Reset Architecture tracks the divergence between an agent's initial assumptions and its acquired knowledge. By employing a "council of agents" to facilitate ritualized dissent, this framework evaluates when new information warrants abandoning the current path. Because AI agents do not suffer from human ego or the sunk cost fallacy, triggering a complete reset—presenting a newly informed prompt to a fresh agent—becomes a viable and cost-effective mechanism. This capability is especially critical for high-stakes, regulated domains like fintech, aviation, and healthcare, where automated decision-making carries real-world risks.

The rise of vibe coding and agent-assisted development is shifting the engineering landscape from rote recall of syntax to architectural recognition. While AI lowers the barrier to entry for rapid prototyping, building safe, scalable software still requires deep domain expertise. Ultimately, engineering leaders must prioritize transparent chain-of-thought logging and responsible AI practices, ensuring that complex agentic workflows remain adaptable, safe, and easily course-corrected before deploying them in critical environments.

**Keywords:** agentic workflows, strategic reset architecture, long-running agent tasks, agile vs waterfall development, AI sunk cost fallacy, vibe coding, responsible AI compliance, chain of thought logging, LLM hallucination mitigation, ritualized dissent, council of agents pattern, complexity theory in AI, software architecture recognition, agentic memory systems, autonomous decision making

## Chapters

1. **Background and career journey in regulated software systems** (00:52) — Transitioning from highly regulated environments to fintech provided a fresh perspective on quality and compliance in software engineering.
1. **How AI agents replicate waterfall instead of agile development** (02:36) — Current agentic workflows execute long-running tasks linearly rather than continually revisiting underlying strategic assumptions.
1. **The limitations of vibe coding and agentic strategy iteration** (05:43) — Rapidly generating and discarding code highlights the failure of current agents to fundamentally rethink flawed initial strategies.
1. **Designing AI agents to exercise doubt and evaluate risks** (08:45) — High-stakes applications require autonomous systems that log their chain of thought and recognize when to pause and reevaluate their approach.
1. **Triggering strategic resets for long-running autonomous tasks** (16:53) — Tracking initial assumptions against discovered knowledge enables agents to restart complex tasks with better-informed baseline strategies.
1. **Overcoming the sunk cost fallacy in software generation** (23:02) — Autonomous agents can discard failed work and restart tasks without experiencing the frustration or ego constraints of human developers.
1. **How AI tools alter developer hiring and architectural design** (27:18) — Lowering the barrier to entry for coding shifts the engineering focus toward architectural expertise and problem recognition over syntax recall.
1. **Publishing research and speaking at tech conferences** (33:52) — Reengaging with the developer community involves sharing new frameworks and proving concepts within live startup environments.

## 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")
- [The evolution of AI programming and agentic workflows](https://www.wearedevelopers.com/videos/100032-under-the-hood-of-building-on-lovable) (from "Under the Hood of Building on Lovable")
- [Navigating developer bottlenecks and human accountability](https://www.wearedevelopers.com/videos/100265-fireside-chat-in-conversation-with-werner-vogels-cto-of-amazon-com) (from "Fireside Chat - In conversation with Werner Vogels, CTO of Amazon.com")
- [Transitioning software engineering teams to AI-native development workflows](https://www.wearedevelopers.com/videos/100087-ai-ready-what-enterprise-transformation-actually-takes) (from "AI-Ready? What Enterprise Transformation Actually Takes")
- [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")
- [Preserving engineering fundamentals in agentic development](https://www.wearedevelopers.com/videos/1897-agents-version-control-and-bunnies-daniel-siegl-david-payr) (from "Agents, Version Control and Bunnies - Daniel Siegl & David Payr")

## 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)
- [Exploring AI: Opportunities and Risks for Developers](https://www.wearedevelopers.com/magazine/522-exploring-ai-opportunities-and-risks-for-developers)
- [Never delegate the understanding](https://www.wearedevelopers.com/magazine/749-never-delegate-the-understanding)
- [The Overflow: AI and Agentic Coding](https://www.wearedevelopers.com/magazine/721-the-overflow-ai-and-agentic-coding)

## Related Jobs

- [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**
- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [Staff Software Engineer, Copilot Experiences](https://www.wearedevelopers.com/jobs/ext/164361-staff-software-engineer-copilot-experiences) at **GitHub**
- [Tribe Lead - ( Software) Engineering Centre of Excllence](https://www.wearedevelopers.com/jobs/ext/1475530-tribe-lead-software-engineering-centre-of-excllence) at **SD Worx**
- [Principal Field Architect - AI Agents](https://www.wearedevelopers.com/jobs/ext/1442858-principal-field-architect-ai-agents) at **Twilio**
- [Twilio's next Senior Principal Field Architect - AI Agents](https://www.wearedevelopers.com/jobs/ext/1487390-twilio-s-next-senior-principal-field-architect-ai-agents) at **Twilio**