The Great Enterprise Refactoring: What Organizations Need to Succeed in the Agent-First Transformation

"We believe that artificial intelligence is akin to the industrial revolution. In particular, we want to call out three points in time of the industrial revolution. The invention of the steam engine, which kicked it all off. the first factory system which put all the requisite parts under the same roof and finally the first factory assembly line as we know it today."

— Sequoia Capital, The Cognitive Revolution

The AI induced refactoring of enterprise software is accelerating, and the organizations that succeed will be those that master four critical capabilities. Here's what separates early winners from the people still struggling to move beyond pilot programs in the cognitive revolution.

Sequoia Capital's latest thesis reveals a profound insight: we're at the steam engine moment of enterprise AI. Large Language Models have reached peak capability—our foundational technology exists—but we're still decades away from the "factory assembly line" equivalent that will unlock the full $10 trillion services automation opportunity. The companies that succeed will master not just better AI models, but the organizational capabilities needed for continuous enterprise transformation.

The Refactoring Reality: Agent-First Success Patterns

Enterprise software integration is undergoing its most fundamental transformation since the move to REST APIs. To materialize this AI-induced refactoring, organizations must rapidly engage in planning, rethinking mindsets, and unlearning traditional approaches while relearning agent-first principles. The early winners are revealing clear success patterns.

The evidence of transformation is mounting:

• Four major protocols have emerged as the foundation for agent-first integrations: MCP for tool integration, ACP for multimodal messaging, A2A for enterprise orchestration, and ANP for decentralized discovery¹
• Real implementations are delivering quantifiable results: JM Family Enterprises achieved 40% time savings in business analysis, ServiceNow automated P1 incident management with multi-agent systems, and Salesforce Agentforce handles 45,000+ conversations weekly with 85% autonomous resolution²
• Companies implementing agent-first architectures achieve 1.5x higher revenue growth and 1.6x greater shareholder returns³

But here's what enables success: winning organizations develop four core capabilities that allow them to decompose monolithic systems, design state management for persistent agent contexts, implement security models for agent delegation chains, and most importantly, engage all stakeholders in the transformation process.

"We've noticed that work is moving from us having minimal leverage on a task and 100% certainty of the outcome to 100 plus% leverage on the task and way less certainty on the exact manifestation of the outcome."

— Sequoia Capital on the leverage vs. uncertainty trade-off

Four Critical Capabilities for Transformation Success

Our research reveals four essential capabilities that separate successful transformations from stalled pilots:

1. Cross-Functional Architecture Literacy

Successful organizations democratize architectural understanding across all stakeholders. Business analysts learn to translate workflows into agent orchestration requirements. Product owners develop visibility into integration complexity and can make informed trade-off decisions. Domain experts participate in architectural decisions with confidence, contributing process knowledge while understanding technical implications.

Success markers:

• 93% of Gen Z workers use two or more AI tools weekly⁴, and winning organizations leverage this native comfort with AI systems
• Cross-functional teams can independently assess agent-first refactoring opportunities
• Architectural decisions include business context and stakeholder input from the start

2. Rapid Learning and Adaptation Culture

The velocity of change demands organizational agility in unlearning traditional approaches and adopting new paradigms. Winning companies create systematic approaches for continuous learning, experimentation, and knowledge sharing across teams.

Key behaviors:

• Regular "unlearning sessions" where teams examine outdated assumptions about system design
• Cross-role pairing programs that build shared understanding between technical and business stakeholders
• Fail-fast experimentation frameworks that encourage rapid iteration on agent-first approaches

3. Stakeholder-Inclusive Planning Processes

Unlike traditional top-down IT initiatives, agent-first transformations succeed when they include all affected stakeholders in planning and decision-making. This ensures that technical architectures align with business processes and user needs.

Implementation strategies:

• Joint planning sessions between technical architects and business process owners
• Regular architecture reviews that include non-technical stakeholders with accessible explanations
• Feedback loops that capture business impact throughout the transformation process

4. Accessible Architecture Intelligence Tools

Organizations need platforms that make architectural expertise available to all stakeholders, not just technical specialists. This enables informed decision-making throughout the refactoring process and ensures business context informs technical choices.

Essential capabilities:

• Visual tools that translate technical architectures into business-understandable formats
• Impact analysis systems that show how technical changes affect business processes
• Decision-support platforms that provide architectural guidance to non-technical stakeholders

"AI linked to a fourfold increase in productivity growth and 56% wage premium, while jobs grow even in the most easily automated roles."

— PwC Global AI Jobs Barometer, 2025

The Historical Parallel: Learning from Industrial Transformation

The Industrial Revolution provides a clear success pattern. The companies that became household names—Westinghouse, Carnegie, Rockefeller—weren't necessarily the inventors of steam technology. They were the organizations that figured out how to systematically apply foundational technology across industries through organizational capabilities.

"The question then becomes who is the John Rockefeller, the Andrew Carnegie, the Westinghouse and the Wedgewood of this cognitive revolution. We believe that it is the startups of today that are carrying out this specialization imperative."

— Sequoia Capital

Today's equivalent success requires developing organizational capabilities for continuous transformation. The rapid unlearning and relearning required for agent-first architectures demands that companies build muscle for ongoing adaptation, not just one-time implementations.

This means:

• Building change management expertise that can guide stakeholders through rapid technology adoption
• Creating feedback systems that capture learnings from each transformation cycle
• Developing partnership strategies with technology providers that can accelerate capability development
• Investing in platforms and tools that make complex technical decisions accessible to business stakeholders

The $10 Trillion Services Opportunity

Sequoia's thesis identifies the $10 trillion US services market where only $20 billion is currently automated by AI. The winners will be organizations that can rapidly build the four critical capabilities needed for continuous transformation.

"This is the $10 trillion US services market of which maybe 20 billion or so today is automated by AI. That is the 10 to the 13th opportunity to not only grow the share of this pie, but grow the pie itself."

— Sequoia Capital

The evidence suggests we're at a unique historical moment. LLMs have reached peak capability—our "steam engine" exists. Agent-to-agent protocols are emerging as industry standards. Early implementations are proving the technology works at scale. But success requires recognizing that the "big refactoring" isn't just a technical challenge—it's a fundamental reorganization of how work gets done that demands new organizational capabilities.

AI agents are delivering 4.8x higher labor productivity growth⁵, but this acceleration only works when organizations develop the capabilities to harness it effectively. The transformation demands investment in people, processes, platforms, and partnerships that enable continuous adaptation.

The companies that master these four capabilities won't just be successful in the current transformation. They'll be the Westinghouse and Carnegie equivalents of the cognitive revolution—organizations that can systematically apply breakthrough technology across industries through superior organizational capabilities.

The $10 trillion opportunity belongs to organizations that can build the capabilities for continuous transformation: cross-functional architecture literacy, rapid learning cultures, inclusive planning processes, and accessible intelligence platforms.

The refactoring has already begun. Success belongs to organizations that invest in the capabilities needed to transform continuously, not just once.

---

Footnotes

1. Based on analysis of Google's Agent2Agent Protocol (A2A), IBM's Agent Communication Protocol (ACP), and emerging MCP standards ↩

2. Enterprise AI implementation case studies from Salesforce, ServiceNow, and JM Family Enterprises ↩

3. PwC Global AI Jobs Barometer, 2025 ↩

4. Workforce AI Adoption Study, 2024 ↩

5. AI Productivity Impact Study, McKinsey Global Institute, 2024 ↩