> Markdown version of [/videos/1540-rethinking-workflows-in-the-agentic-era?t=1414](https://www.wearedevelopers.com/videos/1540-rethinking-workflows-in-the-agentic-era?t=1414). 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).

---

# Rethinking Workflows in the Agentic Era

Single-prompt chatbots are obsolete. The future belongs to multi-step agentic workflows. Learn how to orchestrate specialized AI agents using deterministic metrics and iterative feedback loops.

- **Speakers:** [Eric Jadi](https://www.wearedevelopers.com/@eric-jadi), [Rene Pajta](https://www.wearedevelopers.com/@rene-pajta)
- **Event:** World Congress 2025
- **Published:** August 20, 2025
- **Duration:** 27:07
- **URL:** https://www.wearedevelopers.com/videos/1540-rethinking-workflows-in-the-agentic-era

## Summary

Transitioning from simple, single-prompt chatbot use cases to complex, multi-step scenarios marks the evolution into the agentic era. The true value of large language models lies in automating intricate, non-time-critical workflows where accuracy and quality are more critical than speed. By deeply analyzing human processes, organizations can decompose tasks into injectable units with clear input/output contracts, assigning them to specialized AI agents. This approach shifts the paradigm from one-shot QA to iterative, goal-driven loops designed to handle rigorous evaluations.

Successfully building agent systems requires isolating and defining agent roles, equipping them with the right tools, and orchestrating their interactions through linear chains or dynamic project managers. Pairing a generator agent with a reviewer agent naturally creates a feedback loop that dramatically improves output quality. Engineers can optimize system execution by utilizing smaller language models for routing decisions and leveraging deterministic metrics—such as code compilers or test coverage tools—instead of using an LLM as a subjective judge. Integrating human-in-the-loop checkpoints allows for safe intervention, such as approving logic or architectural budgets, while guardrails prevent runaway costs and infinite recursion.

Real-world applications showcase the flexibility of agent-to-agent architectures. In cloud infrastructure design, specialized agents can query documentation, utilize pricing APIs, and collaboratively estimate Azure architecture details while seamlessly calling a human for verbal budget approvals. Similarly, in software development, agentic workflows can analyze source code, write pseudocode, iteratively review logic, and execute unit tests in a secure environment. Designing these systems with built-in KPIs guarantees that autonomous agents can relentlessly course-correct until they hit target metrics locally before relying entirely on human oversight.

**Keywords:** agentic workflows, multi-agent orchestration, human-in-the-loop AI, generator-reviewer pattern, automated unit test generation, cloud architecture cost estimation, LLM routing models, AI cost optimization guardrails, autonomous project manager agents, code coverage KPIs, autogen framework, agent-to-agent protocols, deterministic evaluation metrics, AI task decomposition, pseudo-code generation

## Chapters

1. **Transitioning beyond single-task AI chat integrations** (00:00) — Moving from simple one-shot queries to complex multi-step workflows provides higher accuracy solutions for enterprise bottlenecks.
1. **Identifying complex workflows for multi-step AI automation** (02:02) — Iterating over problems like backend code migration and cloud infrastructure requires a strategic multi-step design mindset.
1. **Selecting and evaluating candidate workflows for agents** (05:20) — Choosing asynchronous processes where accuracy matters more than speed enables objective scoring and continuous iterative improvement.
1. **Mapping legacy processes to autonomous agent interactions** (09:18) — Decomposing human workflows into single-responsibility programmatic roles clarifies necessary tool requirements and multi-agent decision branches.
1. **Implementing, evaluating, and scaling multi-agent architectures** (13:03) — Isolating agent prompt testing prepares the overall workflow for parallel production execution and strict budget guardrails.
1. **Building an agentic cloud architecture and cost estimator** (16:18) — Specialized autonomous agents orchestrate pricing APIs and system documentation to dynamically validate infrastructure requirements with human supervisors.
1. **Iterating automated unit test generation with coverage metrics** (23:34) — Generating intermediate pseudocode before compiling automated test cases establishes deterministic feedback loops independent of subjective language models.

## Related Moments

- [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")
- [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")
- [Empowering automated workflows with agentic AI models](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")
- [Adapting team structures and agile workflows for agentic tools](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")
- [Addressing the gap between coding assistants and complex workflows](https://www.wearedevelopers.com/videos/100266-ai-won-t-fix-your-engineering-culture) (from "AI Won't Fix Your Engineering Culture")
- [Shifting developer workloads and realistic AI productivity gains](https://www.wearedevelopers.com/videos/1830-wearedevelopers-live-speculaitions) (from "WeAreDevelopers LIVE - SpeculAItions")

## 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)
- [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)
- [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)

## Related Jobs

- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [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**
- [Principal Field Architect - AI Agents](https://www.wearedevelopers.com/jobs/ext/1442858-principal-field-architect-ai-agents) at **Twilio**
- [Staff Software Engineer, Copilot Experiences](https://www.wearedevelopers.com/jobs/ext/164361-staff-software-engineer-copilot-experiences) at **GitHub**
- [AI Operations Manager (all genders)](https://www.wearedevelopers.com/jobs/48263-ai-operations-manager-all-genders) at **envelio**
- [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**