AI agent for cloud migration: How we automate VMware assessment for AWS

When Broadcom acquired VMware, it didn’t just change licensing. It changed the entire game for organizations managing large vSphere ecosystems. Today, IT leaders need answers to new questions:

• How much will my infrastructure cost in AWS?
• Which workloads actually need the resources they’ve been allocated?
• How do I optimize licensing during the transition?
• How can I do this in months instead of years?

Teams approaching this shift usually start with a structured VMware to AWS migration assessment to understand costs, risks, and the fastest path forward.

But the problem isn’t the questions themselves. The problem is how organizations are solving them today. In this article, we uncover how an AI-driven migration assessment platform helps teams analyze VMware environments, identify licensing and sizing risks, and turn raw infrastructure data into AWS migration recommendations.

The old model: “Heroic сonsulting”

Most migrations still follow a pattern that has barely changed in the last 15 years:

1. One or two engineers manually export data from vCenter
2. An audit happens once a year (or when “things get really hot”)
3. Results end up in spreadsheets
4. Someone manually calculates licensing and costs
5. Everything gets redone from scratch when something changes

This works for 50 VMs. It exhausts teams for 500. It’s practically impossible for 5,000. And the worst part? You never know how accurate your recommendations are until you land in AWS and the bills start arriving.

New paradigm: AI-driven, continuous migration

The emergence of AI fundamentally changes this. Instead of “once a year,” you can have:

• Automated discovery – a system that reviews your infrastructure every week
• Intelligent recommendations – AI analyzes not averages, but real peak loads (P95)
• Licensing intelligence – the system automatically identifies which VMs have licensing constraints
• Transparent calculations – every recommendation is backed by data, not assumptions
• Continuous optimization – what you launched a month ago performs better today

In practice, this assessment layer works best as part of broader AWS migration services that cover planning, execution, and post-migration optimization.

This isn’t just “AI that writes a report.” These are specialized agents that rarely communicate, collect data, analyze it, and output structured recommendations. This is a migration factory.

Introducing: OLA — Automated Migration Assessment platform

This is exactly what we built with OLA (Optimization & Licensing Assessment). OLA is not just another VM analysis tool. It gives infrastructure teams an assessment-first starting point before they move into execution, funding discussions, and the wider VMware to AWS migration process. It’s a comprehensive platform that understands the full context of your VMware ecosystem and transforms it into AWS-ready recommendations.

How OLA works under the hood

OLA consists of several key components that work together like a team of specialists:

1. Data collection module (Scanner)

The first agent in the system is the “Scout.” Its job is to safely and completely gather data from your VMware infrastructure.

OLA supports multiple collection methods:

• Direct vCenter API if you have access
• RVTools export – the most popular format in the industry
• Third-party collectors – already installed data collectors
• VMX file uploads – for disaggregated environments

Key feature: The Scout doesn’t just “pull data.” It:

• Normalizes all data into a unified format
• Identifies the operating system on each VM
• Identifies licensing risks (Windows/SQL Server)
• Groups VMs by logical criteria (database servers, web servers, etc.)
• Anonymizes data before any AI analysis

2. Performance analysis module (Analyst)

This is where the magic happens. Unlike traditional tools, OLA doesn’t calculate “average” and say: “You allocated 16 vCPU, let’s give you 8 vCPU in AWS.” Instead, OLA:

Analyzes the 95th percentile (P95)

• What this means: “What’s the maximum power a VM uses during a normal business day (without rare spikes)?”
• Result: Properly sized instances that serve real workload without excessive reserves

Distinguishes “allocated” from “used”

• VMware is often configured with resource guarantees that VMs never actually use
• OLA identifies this “hidden reserve” and shows how much a VM will really cost in AWS

Determines workload classification

• The system automatically recognizes workload types:
  • Enterprise Apps (SQL, Oracle, SAP)
  • Web/API (Apache, IIS, Nginx)
  • Desktop Services (VDI, RDP servers)
  • Data Centers (Hadoop, Kafka, databases)
• Each category receives tailored recommendations

3. Licensing assessment module (Licensing specialist)

This is the most frequently underestimated part of migration.

Most tools say: “Here are 50 VMs, here are 50 AWS instances.”

OLA understands the nuances:

Identifies Windows/SQL VMs and their constraints

• Which VMs have Windows licenses? Which have SQL Server?
• Can I use BYOL (Bring Your Own License) on AWS?
• Will extending licensing to high-core AWS instances be cost-effective?

Calculates licensing costs

• Determines how much licensing will cost on AWS
• Compares: it costs X on-premises, will be Y on AWS
• Creates a list of “quick wins” – VMs where migration will save on licensing

Identifies risks

• “You had a very powerful VM in VMware because Windows licensing was cheaper.” – on AWS, this will be expensive
• “SQL Server on a 64-core machine – this should be split on AWS.”
• “This VM may have licensing restrictions when migrating.”

4. Reporting and recommendations module (Report orchestrator)

This is where users interact with OLA daily.

Each report:

Has a structured dashboard

• Total number of VMs
• Classification by type (Windows, Linux, etc.)
• Estimated savings/costs during migration
• Licensing findings
• Risks and recommendations

Includes detailed recommendations for each VM

• Current configuration
• Recommended AWS instance type
• P95-based resource estimate
• Estimated monthly costs
• Licensing constraints (if any)
• Confidence level of recommendation (100%, 85%, 50%?)

Wave planning

• OLA automatically groups VMs into logical groups (waves)
• First wave can be “quick wins” – VMs that are cheaper on AWS
• Strategic wave – complex workloads requiring more planning

5. AI recommendations module (Analysis with Claude/Bedrock)

OLA goes beyond just numbers. The system uses AI to:

Generate contextual recommendations

• Based on patterns in your infrastructure, AI suggests the optimal strategy
• “You have 20 old VMs that are barely being used – these can be shut down before migration.”
• “These 5 SQL servers could be consolidated on AWS RDS.”

Analyze dependencies

• AI identifies dependencies between VMs
• Plans migration order
• Identifies potential conflict points

Analyze security

• What security configuration is needed in AWS?
• Are any VMs open to the internet in VMware? Will this be a risk on AWS?

Practical OLA features

Data upload and scanning

1. Upload RVTools export or VMX files

2. OLA automatically parses files

3. System identifies:

• OS on each VM
• Installed software (recognizes SQL, Oracle, etc.)
• Network configuration
• Disk and memory volume

Classification and grouping

OLA creates transparent groupings:

• High priority (critical systems requiring special attention)
• Medium priority (typical enterprise apps)
• Low priority (legacy systems, exclusion candidates)
• Risk (systems with licensing or technical risks)

Reports and export

You can export reports in:

• HTML (beautiful presentation)
• Excel (for analysis and manipulation)
• JSON (for integration with your systems)
• AWS Migration Evaluator format (for AWS MAP)

Dashboard and monitoring

Constantly updated dashboard showing:

• Overall analysis progress
• Identified risks
• Estimated savings/costs
• Recommendations requiring action

Real scenario: What migration with OLA looks like

Imagine a typical IT company with 200 VMs. Previously, the process looked like this:

If you want to see how this assessment fits into the broader delivery sequence, our VMware-to-AWS migration playbook breaks down the low-risk path from discovery to cutover.

Key advantages of OLA

This is not a replacement for architects

Important: OLA does not replace people. OLA does what machines should do:

• Collect data without errors
• Analyze patterns
• Generate recommendations based on data
• Live with consolidated information

People (you) do what people should do:

• Understand context (“Is this VM really critical?”)
• Make decisions (“How much risk can we take?”)
• Correct recommendations (“We have legacy constraints”)
• Plan migration waves

Result: Your architects are freed from tedious manual work. They can focus on strategy.

Security and compliance

OLA understands that migration data is sensitive.

Principle 1: Your data stays with you

• OLA can operate completely locally
• Data is not sent to the cloud until you decide

Principle 2: Anonymization

• Before any AI analysis, OLA anonymizes:
  • VM names to generic identifiers
  • IP addresses to generic identifiers
  • Internal domains to generic domains
• Anonymization is deterministic (one VM always maps the same way)

Principle 3: Audit everything

• Each analysis is a separate artifact with a record
• Data versioning – you can see how recommendations changed
• SHA hashes of inputs – for data integrity verification

Practical first steps

If you’re ready to start:

1. Prepare your data

Upload to OLA:

• RVTools export (most popular format)
• Or individual VMX files
• Or direct vCenter API access

2. Run initial scan

OLA automatically:

• Parses files
• Classifies VMs by type
• Identifies licensing constraints
• Prepares initial recommendations

3. Review dashboard and reports

• Overall statistics
• Identified risks and opportunities
• Wave plan suggestions

4. Export and share

Create reports for:

• CFO (cost estimates)
• AWS team (for MAP)
• IT architects (for details)

5. Update regularly

Monthly or after changes:

• Reload new data
• OLA shows what changed
• Update recommendations

Conclusion: Migration is now engineering, not consulting

The old way: migration is a project you do once every three years. The new way: migration is an engineering problem you solve methodically. OLA gives you the tools for the “new way”:

• Data you can defend to the CFO
• Recommendations that improve monthly
• A process that scales with your infrastructure
• Peace of mind knowing nothing was missed

As Peter Drucker said: “What gets measured gets improved.” OLA gives you the ability to measure every VM, every dependency, every licensing risk. And once you measure it, optimization becomes possible.

Ready to start?

The first step is simply a test scan:

1. Export data from vCenter or upload an RVTools file
2. Upload to OLA
3. See what the system finds in 30 minutes

Many companies discover:

• 30-40% resource surplus that can be optimized
• 10-15% VMs that can be eliminated entirely
• USD 50K-500K annual savings (depending on size)

All this becomes visible after the first scan. Don’t let your architects spend time embroidering spreadsheets. Give them OLA. It’s not just a tool. It’s a paradigm shift in how you approach migration. And once the migration plan moves into execution and Day 2 operations, Geniusee also supports that journey with AWS managed services. 

If you are preparing for a VMware-to-AWS migration, OLA can help you assess your environment more quickly and turn infrastructure data into a clearer migration plan. Contact our DevOps experts to get started.