Enterprise AI engineering, governed at scale

Why this matters

AI-enabled software development has helped drive tangible individual productivity gains. However, the challenge is turning those gains into a consistent, governed way for engineers to deliver production software with the quality, traceability and controls large enterprises require. Deloitte’s Ascend Agentic Harness is an AI-powered engineering platform designed to deliver consistent, measurable and enterprise-scale software development. Built on Anthropic’s Claude models, the Ascend Agentic Harness transforms AI-enabled engineering into an organized, repeatable capability for large engineering organizations.

Ascend Agentic Harness overview

Built on Ascend, Deloitte’s AI delivery platform, the Agentic Harness applies Deloitte’s delivery methods, engineering standards, industry experience and quality controls to guide Claude in software development. The platform orchestrates, builds and validates code through repeatable workflows, integrating content from the Deloitte Agentic Exchange—a certified library of reusable agents, scripts and assets. Human and automated checkpoints facilitate security, quality and full requirement-to-code traceability across three phases:

• Orchestrate: Plans work, breaks down tasks, and sets agent guardrails.
• Build: Autonomous agents generate and assemble code within prescribed guidelines.
• Validate: Automated and human reviews confirm delivery meets quality, security and compliance requirements.

Through features such as domain knowledge briefs, reusable script logic, token optimization, structured decision gates and security checkpoints, the harness enables governed development for new and legacy systems alike.

Key benefits

• Enterprise consistency: Provides governed, traceable software delivery with structured checkpoints, from orchestration to deployment.
• Supply chain security: Helps remediate vulnerabilities across custom, commercial and open-source software layers.
• Operational focus: Prioritizes fixes based on real business risk, not just raw volume, keeping teams focused on what matters most.
• Engineering productivity: Automates patch generation, validation and compliance, allowing engineers to focus on architecture and innovation.
• Modular flexibility: Supports a full lifecycle of needs, from software visibility through to risk management and compliance reporting.
• Flexible deployment: Available as a fully managed service, client-owned build, or on-site engineering augmentation to fit diverse enterprise requirements.

The Ascend Agentic Harness in Action

Deployed to 35,000 Deloitte engineers, the Ascend Agentic Harness is currently used in complex client engagements—including data and application modernization for leading insurers, banks, health care payers, manufacturers and telecommunications organizations. It helps teams advance projects at greater speed and scale while maintaining rigorous oversight and compliance.

Recently, Deloitte engineers used the Ascend Agentic Harness to reverse engineer legacy data platforms transforming them into modern platform-based data products for a top regional banking institution. Similarly, it was used at a large insurance organization to drive data platform modernization simplification with needed lineage, transparency, data quality, catalogue and generate data products, while reducing overall data platform maintenance efforts.

Forward deployed engineers (FDEs): Embedded support for modernization

The next generation of processes and workflows demands reinvention for agent-driven autonomy, and clients need help architecting and integrating AI-enabled platforms into their enterprise stack. To meet this need, Deloitte is further scaling teams of FDEs. These engineers bring vertical experience and work directly alongside client technology and business teams—embedded in the client environment—to modernize legacy IT systems and reimagine business processes using Claude.

“The Ascend™ Agentic Harness moves the engineering community beyond AI experimentation into governed, enterprise-scale delivery—enabling clients to advance with confidence.”—Parth Patwari, Principal, US AI and Engineering Leader, Deloitte Consulting LLP

The model pairs Deloitte's engineering and industry experience with hands-on, in-context delivery, helping clients architect and integrate AI-enabled platforms into their enterprise stack while moving faster from working prototypes to solutions in production.