Many AI strategies discover the same problem.
The model is capable.
The use case is valuable.
The legacy environment cannot support it.
The information is buried in a system nobody wants to touch.
The integration is brittle.
The business rules exist in code, spreadsheets, and people's memory.
The user experience cannot support review, recommendation, or intelligent action.
A small change takes months.
The company may describe this as a data problem, an integration problem, or technical debt.
It is also an AI strategy problem.
Modernization is often the first step of an AI strategy.
AI exposes what the legacy system was already hiding
A legacy platform can continue operating for years while creating invisible drag.
Employees learn the workarounds.
IT learns which areas are dangerous to change.
Reports are produced through special queries.
Integrations are patched.
Business rules are copied into spreadsheets.
Customers accept an outdated experience because there is no alternative.
Then the company tries to introduce AI.
AI needs access to information.
It needs clear actions.
It needs reliable interfaces.
It needs visible rules.
It needs a way for people to review, approve, correct, and understand the result.
The AI initiative does not create the weakness.
It reveals it.
Four constraints block AI repeatedly
1. Inaccessible data
The information may exist, but not in a form the new capability can use reliably.
It may be trapped in proprietary tables, inconsistent records, documents, exports, or reports designed for humans.
The solution is not always a giant data-platform program.
It is often a focused effort to make the information required by one valuable workflow available, trusted, and permission-aware.
2. Brittle integrations
An AI-enabled experience often needs to read from and act across several systems.
If those systems have unclear APIs, one-off connections, or fragile point-to-point integrations, every new capability becomes risky.
Modernization can create clearer services and boundaries so products, workflows, and AI can interact with the business reliably.
3. Hidden business logic
Important decisions may be encoded in old application code, undocumented rules, report formulas, or employee workarounds.
AI cannot be governed around rules the organization cannot see.
Modernization makes the logic explicit enough to test, explain, and improve.
4. An outdated operating experience
A legacy interface was designed for the old process.
It may not support recommendations, generated work, human review, exception handling, or adaptive workflows.
Sometimes the company does not need a smarter feature inside the old interface.
It needs a new experience around the work.
Modernize with AI and for AI
Modernize + AI contains two distinct ideas.
Use AI to rebuild faster
AI can accelerate parts of the modernization process:
- Understanding legacy code
- Reconstructing documentation
- Identifying dependencies
- Generating repetitive components
- Expanding test coverage
- Mapping data
- Preparing migration work
- Supporting quality assurance
Experienced engineers must still own the architecture, security, quality, and production decisions.
AI changes the amount of manual engineering work required.
It can also improve coverage and speed.
Rebuild so AI can create value afterward
The resulting platform should not simply be a cleaner version of the past.
It should make the next capabilities easier to introduce.
That may require:
- Accessible, well-structured data
- Clear services and APIs
- Modern identity and permissions
- Visible business rules
- Event and workflow state
- Auditability
- A flexible user experience
- Monitoring and evaluation
- Architecture that can evolve
The goal is not simply to come out current.
It is to come out more capable.
Not every system needs to be replaced
An AI strategy does not require the company to modernize everything.
That can become another reason not to begin.
The better question is:
Which system, integration, data constraint, or experience is preventing the highest-value work from changing?
Modernize that constraint first.
A company may preserve the core system of record while building a new interface around it.
It may replace one module.
It may expose a new service.
It may move one workflow onto a modern foundation.
It may rebuild the customer portal before the back-office system.
It may create a parallel modern platform and migrate in phases.
Modernization should be guided by business value and operating risk, not by a desire to make every part of the architecture beautiful.
Phased delivery changes the risk
Traditional modernization programs often ask the business to fund years of work before users see meaningful value.
That is a weak operating model even when the technical plan is sound.
A phased approach creates useful releases along the way.
At Clear Packaging, VBT first launched a modern client portal on a replicated database. Customers gained faster self-service access while live laboratory operations remained protected.
Then the back-office platform was rebuilt module by module, with parallel runs and complete historical-data migration.
Report performance moved from minutes to seconds. Reporting flexibility increased tenfold. Every phase shipped, and the laboratory continued operating.
The business did not have to wait for the final rewrite to learn whether the modernization was working.
Each release earned the next.
At ID Fund, the investor and back-office platforms were rebuilt over roughly ten months. Every account and record migrated with zero operational downtime.
The new architecture improved the current experience and created a stronger foundation for analytics, continued product development, and future AI integration.
Modernization created immediate value and future option value.
Do not bolt AI onto the constraint
There is pressure to show AI progress quickly.
That can lead teams to build around the legacy limitation rather than resolve it.
They export the data manually.
They create a side database.
They build an AI interface with no reliable write-back.
They duplicate business rules.
They create another temporary workflow.
Sometimes that is acceptable for a pilot.
It becomes dangerous when the workaround quietly becomes the production architecture.
The question is not whether the company can connect AI to the old system somehow.
It is whether the resulting capability will be reliable, maintainable, governable, and capable of improving.
How to choose the first modernization slice
Look for the intersection of:
- Business constraint
- Technical constraint
- Practical independence
- Measurable value
A strong first slice may be:
- The portal preventing customer self-service
- The reporting layer creating management delay
- The integration preventing one high-value workflow
- The product-data foundation required by recommendations
- The back-office module creating manual workarounds
- The service boundary required by several future capabilities
The first slice should be capable of launching independently enough to prove value.
It should also move the architecture in the direction the company needs.
Modernization is not the tax before innovation
Companies sometimes view modernization as the expensive cleanup required before the interesting AI work can begin.
That framing undersells the opportunity.
Modernization can itself be the first AI initiative.
AI can accelerate the rebuild.
The new platform can remove manual work.
The redesigned experience can support intelligent assistance.
The data and services can enable future products.
The release cadence can create immediate operating improvement.
The company does not have to choose between fixing the past and building the future.
It can use one effort to do both.
The leadership question to ask
When an AI initiative stalls, do not ask only:
Which model or vendor should we try next?
Ask:
What is the underlying system preventing this work or product from becoming possible?
Then ask:
What is the smallest modernization step that creates immediate value and changes what the company can build afterward?
Modernization is often the first step of an AI strategy.
Not because every legacy system must be replaced.
Because the systems beneath the work determine what intelligence can become real.
