AWS’s Agentic AI Modernization Strategy to Reset Legacy IT

Agentic AI modernization took center stage across AWS re:Invent keynotes, briefings, and product launches this year. Systems that reason, act, learn, and execute across workflows are now positioned as the next phase of enterprise automation. But inside most organizations, the same question still appears in every boardroom: What do these announcements actually mean for the systems keeping businesses running daily?

AWS used re Invent 2025 to answer that question directly. Rather than offering far future demos, AWS focused on something enterprise leaders know well technical debt. Years of legacy systems, mainframes, Windows estates, VMware environments, and brittle integrations slow every transformation effort. AWS presented agentic AI modernization as the toolset that finally makes large-scale modernization possible at speed.

Across AWS Transform, Nova, Nova Forge, Bedrock AgentCore, Frontier Agents, AI Factories, and its partner ecosystem, AWS delivered one connected narrative: reset the past, automate the present, and use agentic systems to reshape how enterprise software is built, secured, and operated.

Tech Debt and the Need for Agentic AI Modernization

AWS placed technical debt at the core of its story. A typical organization spends more than 30% of team time on manual modernization work effort required to keep systems alive but which drains momentum from innovation.

AWS Transform has already analyzed more than 1.1 billion lines of code and removed over 810,000 hours of manual work. Early adopters modernized systems four times faster than traditional approaches.

AWS argues that aging systems are the hidden barrier limiting enterprise AI. Even advanced automation stacks struggle when connected to brittle architectures built decades ago. This shifts the conversation away from AI novelty and toward infrastructure reality.

In a symbolic move, AWS even staged a demolition of decommissioned servers showing that legacy systems no longer need to remain frozen in place. Agentic AI modernization becomes the demolition tool, reconstruction tool, and long-term operating model.

AWS Transform Becomes the Heart of Agentic AI Modernization

The expansion of AWS Transform became the week’s centerpiece. AWS described it as the first agentic AI service built explicitly for massive application and code modernization efforts.

Transform now supports:

• Windows .NET

• SQL Server

• UIs and operating systems

• Mainframes

• VMware

• APIs

• Legacy runtimes

• Organization-specific languages

AWS claims Transform can accelerate complete Windows modernization by up to five times and cut maintenance costs by up to 70%.

Customer examples prove the point:

• Air Canada modernized thousands of Lambda functions in days, cutting cost/time by 80%.

• QAD reduced upgrade cycles from two weeks to three days.

• Thomson Reuters now migrates 1.5 million lines of code monthly, reducing technical debt by half.

Madhu Parthasarathy, general manager of AgentCore, summarized this shift clearly: “Several thousands of hours reduced into weeks.” This is where agentic AI modernization breaks from traditional automation Transform captures outcomes and improves with every project.

Modernization stops being a decade-long initiative and becomes continuous operational work.

Mainframes and VMware Become Targets for Agentic AI Modernization

AWS extended Transform into mainframe and VMware estates two environments that historically resisted automation.

New mainframe agents now automate activity analysis, business rule extraction, documentation, and test planning. These previously manual steps often stalled migrations and created huge project risks.

AWS also introduced capabilities to automate VMware migrations, including support for Cisco ACI, FortiGate, and Palo Alto Networks. A new on-premises discovery tool improves estate visibility for security teams.

For enterprises, agentic AI modernization now handles dependency mapping, documentation, and planning tasks once reliant on months of manual labor.

Frontier Agents and DevOps Reinvented Through Agentic AI Modernization

AWS introduced three Frontier Agents:

• Kiro – autonomous development agent

• AWS Security Agent – supports secure application design and testing

• AWS DevOps Agent – focuses on reliability, prevention, and incident response

These agents operate with context awareness and minimal human oversight. Engineers now guide intent, while agentic systems execute the repetitive layers.

This marks a structural shift in how software teams function. Agentic AI modernization becomes a co-worker rather than a tool.

From Prototypes to Production: The Maturity of Agentic AI Modernization

AWS made one message clear: the industry is crossing the chasm from prototypes to production.

Madhu Parthasarathy shared a striking prediction:

“2026 will be the year when we see many more agents deployed in production at scale.”

Trust not technology is now the primary barrier. This is why Bedrock AgentCore’s new policy layer became a critical announcement.

Policy enforcement enables risk teams to maintain determinism in a non-deterministic system. Combined with episodic memory and evaluations, this pushes agentic AI modernization from experimentation into enterprise-grade reliability.

Multi-Agent Orchestration: The New Architecture for Agentic AI Modernization

AWS outlined a new design pattern: letting reasoning models orchestrate workflows.

Instead of hard-coded logic, models now determine steps dynamically. This is powered by:

• Nova models for reasoning, multimodal input, and code execution

• Nova Forge for custom training using proprietary data

• Nova Act for automating browser based workflows like CRM updates and claims processing

Hertz and 1Password already use Nova Act to compress QA cycles from weeks to hours.

This reinforces that agentic AI modernization is not only about updating old systems—it’s about re-architecting enterprise automation.

Partners Become a Force Multiplier for Agentic AI Modernization

AWS partners now generate $7 in services revenue for every $1 of AWS spend. More than 80% offer AI-driven transformation services.

New categories for agentic AI partners and Marketplace updates like multi-vendor bundles signal a future where deployments increasingly run through partner ecosystems.

Readers can explore AWS’s partner programs on the official site.

Learn more about AWS agentic systems in their relevent recap.

Conclusion: The Era of Agentic AI Modernization Begins

Taken together, re:Invent 2025 shows AWS consolidating its AI strategy around agentic AI modernization. Transform handles legacy modernization, Nova supplies reasoning, Bedrock AgentCore ensures trust, and Frontier Agents redefine engineering.

Just as the Gold Rush rewarded those who sold tools rather than mining gold, AWS and its partners aim to supply the virtual picks and shovels for this new enterprise AI era.