AI Dev 26 x SF | João Moura: Building Recurring, Governed, and Embedded Enterprise Workflows

2026-05-22 · Source: DeepLearningAI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Software Development & Engineering · Depth: Intermediate, extended

Summary

CrewAI's internal coding agent, Project Iris, initially faced user resistance but achieved significant adoption, now altering almost half of the company's Pull Requests. This success, alongside other internal applications like automated sales leave-behind material generation, demonstrates the growing utility of AI agents beyond coding. CrewAI aims to accelerate enterprise agent adoption by addressing two key industry forces: the diverse types of agent systems (ad hoc vs. embedded workflows) and the commoditization of building AI solutions. Their approach focuses on reusable building blocks, integrating human-in-the-loop processes via email notifications, and providing zoom-out/zoom-in metrics for performance and cost. The company views agent adoption as a transformational challenge, emphasizing strategy, discovery, and user engagement over purely engineering concerns, and is developing self-improving, long-running, conversational agents.

Key takeaway

For Directors of AI/ML evaluating enterprise agent adoption, recognize this as a transformational challenge beyond pure engineering. Prioritize developing reusable agent components and integrating human-in-the-loop processes early. Focus on establishing clear strategies and fostering user adoption, not just technical deployment. Your success hinges on enabling self-improving, long-running agents and comprehensive metrics, ensuring alignment with organizational goals and accelerating value delivery.

Key insights

Enterprise agent adoption is a transformational problem requiring strategic planning, reusable components, and human-in-the-loop integration for success.

Principles

• Agent adoption is a transformational problem.
• Reusability drives agent value and speed.
• Integrate humans-in-the-loop effectively.

Method

Accelerate agent adoption by building reusable tools, agents, and sharable skills. Integrate human-in-the-loop via email notifications and monitor performance with zoom-out/zoom-in metrics for cost and execution health.

In practice

• Implement internal coding agents (e.g., Project Iris).
• Automate sales leave-behind material generation.
• Encode company decisions into engineer terminals.

Topics

• AI Agents
• Enterprise AI Adoption
• Human-in-the-Loop
• Reusable AI Components
• MLOps Metrics
• Self-improving Agents

Best for: CTO, VP of Engineering/Data, Executive, AI Engineer, MLOps Engineer, Director of AI/ML

Related on AIssential

• How Endava is redesigning software delivery around AI agents — Organizations can achieve AI-native operations by embedding AI agents as the primary problem-solving approach across all workflows.
• #5: AI Workflow Patterns: The Real Unit of AI Adoption in 2026 — AI workflows, not models, are the fundamental unit for organizational improvement and AI adoption.
• Agentic AI: From Hype to Enterprise Deployment, Challenges, Frameworks, and Real ROI in 2026 — Agentic AI is moving from pilots to production, requiring robust governance and integration for measurable enterprise ROI.
• Mendo secures €12M to scale enterprise AI adoption in Europe — The core challenge in enterprise AI is adoption and integration, not just technology implementation.
• Box CEO on AI Agents & Why Enterprise Can't Keep Up | a16z — Enterprise AI adoption is hindered by integration complexities, architectural paralysis, and a mismatch between tech innovation and organizational readiness.
• Anthropic and OpenAI Battle for Enterprise AI — Enterprise AI adoption hinges on effective deployment and workflow integration, not just advanced models.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by DeepLearningAI.