AWS announced today (05/14/2026) the new Agent Builder Toolkit for AWS Transform.
But the announcement makes much more sense when connected to the platform’s recent evolution.
AWS did not launch just another isolated feature.
Step by step, it is building an agent-oriented platform for enterprise modernization.
And honestly, this may be one of the most coherent applications of AI in cloud infrastructure today.
For AWS’s official write-up of today’s Agent Builder Toolkit announcement—scope, capabilities, and product framing—see the What’s New post:
AWS Transform Agent Builder Toolkit — What’s New (AWS)
What was announced today
Today’s primary announcement was the Agent Builder Toolkit.
Customers and partners can now:
- build custom transformation and modernization agents
- share agents across teams
- register agents within the AWS Transform ecosystem
- create specialized workflows for specific modernization scenarios
In practice, AWS Transform is starting to move beyond being just a migration tool and becoming an extensible agent platform.
And that distinction matters.
Because it fundamentally changes the role the platform plays inside enterprise organizations.
The broader context matters more than the isolated announcement
If you look only at today’s release, it can sound like “just another toolkit.”
But when connected to AWS’s recent announcements, the direction becomes much clearer.
AWS is gradually turning enterprise modernization into an agent-oriented workflow.
Automated Landing Zone creation connects directly to this
In April 2026, AWS announced that AWS Transform could automate landing zone creation directly within migration workflows.
🔗 AWS Transform automates landing zone (What’s New)
That announcement did not happen today.
But it is critical for understanding the current direction of AWS Transform.
Because landing zones have always been one of the most operationally expensive parts of enterprise modernization.
This is not simply about “creating AWS accounts.”
It involves an entire layer of:
- governance
- IAM
- Organizations
- Control Tower
- networking
- compliance
- multi-account structures
- observability
Historically, all of this required significant manual work and operational overhead.
Now AWS Transform can:
- understand migration context
- recommend organizational structures
- generate Infrastructure as Code
- export to CloudFormation, CDK, or LZA
In practice, AWS is reducing the repetitive operational work required to prepare enterprise environments before modernization even begins.
And this is where agents actually make sense
This is exactly the type of problem space where agents provide real value.
Because enterprise modernization involves:
- distributed context
- repetitive workflows
- inconsistent architectures
- fragmented documentation
- organizational dependencies
- high cognitive overhead
This is fundamentally different from indiscriminately adding AI to problems that deterministic software already solves well.
In the OverAI article, I argued exactly this:
there is a difference between using AI to solve a structural problem and using AI simply because it became a trend.
And enterprise modernization clearly falls into the first category.
IDE and MCP integration reinforce the same direction
Another important move was AWS Transform integration with:
- Kiro
- Claude
- Cursor
- Codex
It is now possible to start transformations directly from the IDE, monitor execution in the AWS console, and receive results back inside the editor.
Official reference on AWS Transform agents in Kiro, agent plugins, and the AWS Transform MCP server:
AWS Transform agents now available in Kiro, Claude, Cursor, and Codex — What’s New (AWS)
This significantly reduces the separation between:
- development
- cloud infrastructure
- automation
- modernization workflows
- agents
But perhaps the most important part here is MCP.
MCP may be the most strategic part of all this
AWS Transform now also supports MCP (Model Context Protocol).
And that changes quite a bit.
MCP is quickly becoming a standard integration layer between:
- agents
- IDEs
- platforms
- workflows
- tools
In practice, this means AWS Transform can now participate programmatically in larger agent-oriented automation workflows.
And that makes the platform far more composable.
The Agent Builder Toolkit closes the loop
Today’s announcement effectively completes this architecture.
AWS now enables organizations to:
- build specialized agents
- reuse organizational workflows
- share capabilities across teams
- compose agent-oriented pipelines
The pattern is becoming increasingly clear.
Historically:
- tools executed commands
Now:
- agents execute objectives
Before:
- “configure IAM”
- “create a VPC”
- “provision infrastructure”
Now:
- “prepare this organization for migration”
- “modernize this legacy system”
- “transform this architecture”
That difference is significant.
Because enterprise modernization has never been purely about technical execution.
It requires interpretation, context, and organizational coordination.
AWS’s larger strategy is starting to emerge
Looking at AWS’s recent moves:
- Amazon Bedrock AgentCore
- Claude Platform on AWS
- MCP
- AWS Transform
- IDE integrations
- extensible agent platforms
Everything points in the same direction:
AWS is evolving beyond infrastructure delivery and becoming the operational layer for agent-oriented systems.
And honestly, this direction feels much more grounded than a large part of today’s AI hype cycle.
Because here there is clear alignment between:
- real operational problems
- organizational scale
- repetitive workflows
- automation needs
- distributed context
Conclusion
Today’s announcement is not just about creating custom agents.
It reinforces a direction AWS has been building toward over the last several months.
AWS Transform is evolving from a modernization tool into an agent-oriented platform.
And this may be one of the clearest examples of AI being applied where it genuinely makes sense:
- complex workflows
- high organizational context
- enterprise modernization
- cognitive automation
- infrastructure transformation at scale
Not as narrative.
But as real operational capability.