Westpac Intranet Migration AI Tokens: How $12,000

The Westpac intranet migration AI tokens story is one of the clearest real-world examples yet of how generative AI can change the cost and speed of enterprise technology projects.

Westpac has disclosed that an AI-assisted migration of its internal employee intranet consumed about US$12,000 worth of AI tokens. According to iTnews, the bank said a comparable project might traditionally have cost in the low seven figures and taken roughly a year of work.

The project was not a tiny proof of concept.

A Westpac team migrated nearly 7,000 pages and more than 18,000 documents across seven branded intranets into a new consolidated employee experience called MyW. A project leader involved in the migration said the work was completed in about 61 hours, compared with an estimated 12 to 18 months of manual effort for a migration of similar scale.

That is the headline-grabbing part.

The more important question is what the project actually tells us about enterprise AI, cost control, automation, and the limits of simple return-on-investment calculations.

What Was the Westpac Intranet Migration?

Westpac was consolidating multiple internal employee intranets into a new unified environment.

The project involved moving:

almost 7,000 pages
more than 18,000 documents
seven branded intranets
page structures
links
images
brand elements

The final system is known as MyW, Westpac’s new employee intranet.

The project also aimed to improve:

navigation
search
content governance
trusted-information discovery
future personalization
access through Microsoft Copilot

So this was not just copying files from one system to another.

The migration also involved restructuring and improving how information would be found and managed after the move.

How Much Did the Westpac Intranet Migration Cost in AI Tokens?

According to Westpac chief AI officer Dan Jermyn, the project consumed about US$12,000 in AI tokens.

That figure refers to the model-inference cost associated with running AI systems during the migration.

AI platforms commonly charge based on token usage.

A token is a small unit of text processed by a model. Depending on the system, both input and output tokens can contribute to the total cost.

For a large migration, AI may need to process:

page content
metadata
document structures
links
headings
images
classifications
migration rules
error messages
transformation instructions

Each of those interactions consumes model capacity.

The surprising part is not that the project used many tokens.

It is that the reported token bill was still only around US$12,000 compared with the scale of the work being automated.

Why the $12,000 Figure Matters

The Westpac intranet migration AI tokens figure matters because enterprise AI discussions often stay vague.

Companies frequently say AI makes work:

faster
cheaper
more efficient
more productive

But they do not always provide a simple, measurable cost comparison.

Westpac’s example is unusually concrete.

iTnews reported that the bank viewed the migration as something that might traditionally have cost low seven figures.

If that comparison is directionally accurate, the difference between a seven-figure manual project and a US$12,000 AI-token bill is enormous.

However, that does not mean the entire project cost only US$12,000.

The token figure does not automatically include:

employee salaries
engineering time
project management
infrastructure
vendor costs
testing
quality assurance
governance
change management
security review

That distinction is important.

The AI-token bill is one component of total project cost, not the entire economic picture.

How Fast Was the Migration?

A Westpac project leader said nearly 7,000 pages and more than 18,000 documents were migrated in about 61 hours.

The same post said a migration of that scale could normally require 12 to 18 months of manual effort.

That is a dramatic difference.

But it should be interpreted carefully.

The 61-hour figure appears to refer to the migration execution rather than every phase of the project.

Before and after a migration, teams may still need to perform:

planning
mapping
validation
stakeholder coordination
content review
acceptance testing
access-control checks
post-launch support

So the strongest claim is that AI appears to have dramatically accelerated the execution of the migration.

It would be misleading to assume that an entire enterprise transformation project, from first meeting to final handover, happened in 61 hours.

What Did AI Actually Do?

Public descriptions suggest AI was used to automate large parts of the migration process.

That likely included tasks such as:

interpreting source-page structures
moving content
maintaining links
preserving documents
handling images
mapping content between platforms
automating repetitive migration logic

The project also had to preserve brand identity where necessary while combining seven separate intranets into one experience.

This is where AI can be particularly useful.

Traditional migration work often involves thousands of repetitive decisions that are individually simple but collectively expensive.

AI systems can process and transform content at a scale that would be tedious for human teams.

Humans are still needed to define the rules and check the results.

That division of labor is one of the more realistic models for enterprise AI.

The Role of Westpac’s AI Engineering Strategy

The migration did not happen in isolation.

Westpac has been expanding its broader AI engineering strategy.

In August 2026, the bank announced a strategic partnership with Amp Frontier Corporation, an AI engineering company focused on software-development agents.

Westpac said Amp engineers would work alongside its teams to help accelerate technology delivery and improve digital experiences.

The bank described the broader goal as moving some technology programs from months to weeks.

The intranet migration was cited as an early example of that broader approach.

This makes the project more interesting than a one-off automation experiment.

It is part of a wider attempt to embed AI into engineering workflows.

Westpac Is Also Tracking AI Token Costs

The bank is not treating AI usage as free.

Bloomberg Law reported in July 2026 that Westpac was closely tracking employee AI-token consumption and encouraging staff to use the appropriate model for each task.

That matters because more capable models can also be more expensive.

Using the largest available model for every task would waste money.

A sensible enterprise AI strategy may instead route:

simple tasks to cheaper models
complex reasoning to stronger models
repetitive transformations to specialized systems
sensitive work to approved internal environments

The Westpac intranet migration AI tokens story therefore has two sides.

AI can dramatically reduce some project costs, but organizations still need to monitor AI spending carefully.

Why Model Routing Matters

AI pricing can become expensive at enterprise scale.

Imagine a company processing millions or billions of tokens across:

software development
customer service
data analysis
internal search
document processing
marketing
compliance
fraud detection

Small differences in model cost can become large differences in annual spending.

Westpac’s approach appears to include routing work to different AI models depending on task complexity and cost.

This is a useful lesson for other businesses.

The goal should not simply be:

Use the smartest AI model.

A better question is:

What is the cheapest model that can reliably complete this task?

That is much closer to how enterprises already manage cloud-computing resources.

Why This Project Was a Good AI Use Case

Not every business process is equally suitable for generative AI.

Content migration is a particularly interesting use case because much of the work is repetitive but still requires context.

A traditional script may struggle when thousands of pages contain slightly different:

layouts
headings
links
image structures
metadata
content types

AI can interpret those variations more flexibly than rigid automation.

At the same time, the task has a clear output that can be checked.

A migrated page either:

exists
contains the correct content
preserves the right links
includes the expected files
passes validation

That makes quality assurance easier than in more subjective AI tasks.

The Human Team Still Mattered

The migration should not be interpreted as AI replacing an entire project team.

One of the project leaders emphasized that the outcome still depended on people with technical and domain expertise.

The team included people with skills in areas such as:

SharePoint
engineering
content migration
employee experience
technical architecture
project delivery

AI accelerated the work.

Humans still had to:

define objectives
design the process
solve unusual cases
validate results
make decisions
manage deployment

That distinction matters whenever businesses calculate AI ROI.

The best use of AI is often not removing humans completely.

It is allowing a skilled team to achieve much more with the same amount of time.

Westpac’s Broader AI Push

Westpac has been expanding AI use across the bank.

Its 2026 interim results presentation said Microsoft 365 Copilot access and training had been made available across the workforce, while specialist AI tools were being provided for engineering and analytics teams.

The same presentation highlighted AI use in:

scam detection
fraud monitoring
customer analytics
business insights
transformation programs
productivity

Westpac also said AI-assisted impact assessments for its transformation work had reduced turnaround times from about 10 days to under four days.

That suggests the intranet project is part of a broader push rather than an isolated experiment.

Does the Project Prove AI Has Huge ROI?

Not automatically.

Westpac executives themselves have been careful about making overly broad claims.

iTnews reported that chief data, digital and AI officer Andrew McMullan said isolating the exact contribution of AI is not always straightforward.

This is sensible.

Enterprise projects have many inputs.

If a migration becomes faster, the improvement might come from:

AI
better tooling
experienced engineers
improved project design
cleaner source content
better cloud infrastructure
previous migration knowledge

Separating those effects can be difficult.

The Westpac project offers a strong example of AI-enabled efficiency.

It should not be treated as proof that every AI project will produce the same economics.

The Problem With Comparing Token Cost to Labor Cost

A simple comparison might look like this:

Traditional migration: seven figures
AI tokens: US$12,000

That sounds like the AI made the project almost free.

But the comparison is incomplete.

Token spending and labor spending are not equivalent categories.

The AI system still required:

engineers
supervision
infrastructure
implementation
testing
governance

A better comparison would include the total cost of the AI-assisted project against the total cost of a traditional project.

Westpac has not publicly disclosed every cost component.

So the US$12,000 number should be seen as a useful data point rather than a complete financial statement.

What Other Companies Can Learn

The project provides several practical lessons.

Start with repetitive, measurable work

AI is easier to evaluate when the result can be checked.

Migration is a strong example because outputs are concrete.

Track token costs

AI usage can scale quickly.

Companies should monitor:

tokens consumed
model used
cost per task
failure rate
human-review time

Use the right model for the job

Expensive models should not automatically handle simple work.

Model routing can reduce costs significantly.

Keep humans in the loop

Enterprise content can contain:

sensitive data
compliance requirements
access restrictions
unusual exceptions

Human review remains essential.

Measure total project economics

Token cost alone is not enough.

Organizations should compare:

total labor
AI costs
infrastructure
project duration
rework
support
quality

Could AI Change Enterprise Migrations?

Very likely.

Large organizations constantly migrate:

intranets
content-management systems
document repositories
cloud platforms
databases
customer portals
legacy applications

These projects can be expensive because they contain enormous amounts of semi-structured information.

Generative AI is well suited to interpreting semi-structured content.

That could reduce the need for manually writing thousands of one-off transformation rules.

The Westpac case offers one early example of what that future might look like.

Risks of AI-Assisted Migration

The benefits do not remove the risks.

Incorrect content mapping

AI may classify or move content incorrectly.

Broken links

Large migrations can create links pointing to old or missing locations.

Permissions

Internal intranets may contain restricted information.

Incorrect permission handling could expose confidential content.

Hallucinated changes

Generative models may occasionally produce information not present in the source.

Migration systems should avoid allowing models to invent or rewrite content unless explicitly required.

Auditability

Organizations need records showing:

what was moved
what was changed
what failed
who approved it

This is particularly important in regulated industries such as banking.

Why the Banking Context Matters

Westpac operates in a highly regulated industry.

That makes the project more notable.

Banks have strict requirements around:

security
privacy
access control
data governance
operational resilience

A successful AI-assisted migration in this environment suggests that AI automation can be applied to serious enterprise workflows when governance is designed properly.

It does not mean those controls can be skipped.

Quite the opposite.

The more automation is used, the more important validation and monitoring become.

What Is MyW?

MyW is the new intranet environment created through Westpac’s migration project.

The goal was to bring seven separate branded intranets into one employee experience.

The project also aimed to improve how employees:

navigate information
search internal content
find trusted resources
access content through Copilot
interact with personalized experiences

That makes MyW more than a storage location.

It is intended to become part of Westpac’s broader digital workplace.

Westpac Intranet Migration AI Tokens vs Traditional Delivery

A simplified comparison looks like this:

Metric	AI-Assisted Westpac Project	Traditional Estimate
AI token cost	About US$12,000	Not applicable
Pages	Nearly 7,000	Same scale
Documents	More than 18,000	Same scale
Migration execution	About 61 hours	12–18 months manual effort
Intranets consolidated	7	7
Total project cost	Not publicly disclosed	Reported as low seven figures

The important row is the last one.

We know the AI-token cost.

We do not know the complete cost of the AI-assisted project.

Frequently Asked Questions

What is the Westpac intranet migration AI tokens story?

Westpac disclosed that an AI-assisted employee-intranet migration consumed around US$12,000 worth of AI tokens. The project moved thousands of pages and documents into a new consolidated intranet.

How many pages did Westpac migrate?

A project leader said nearly 7,000 pages were migrated.

How many documents were involved?

More than 18,000 documents were included in the migration.

How long did the migration take?

The migration execution was reported as taking about 61 hours. A comparable manual project was estimated at 12 to 18 months.

Did the whole project cost only US$12,000?

No public source supports that conclusion.

US$12,000 refers to the reported AI-token spend. Total costs such as labor, infrastructure, engineering, testing, and project management were not fully disclosed.

What is MyW?

MyW is Westpac’s new consolidated employee intranet created from seven previously separate branded intranet environments.

Is Westpac using AI elsewhere?

Yes. Westpac has publicly discussed AI use in fraud detection, customer analytics, engineering, Microsoft 365 Copilot, and transformation programs.

Conclusion

The Westpac intranet migration AI tokens story provides a rare concrete look at enterprise AI economics.

The bank spent about US$12,000 in AI-token usage on a migration involving nearly 7,000 pages, more than 18,000 documents, and seven intranet environments.

The migration execution was reported at roughly 61 hours, compared with an estimated 12 to 18 months of manual work.

Those figures are impressive.

But the useful lesson is not that every seven-figure IT project can suddenly be completed for US$12,000.

The better lesson is that AI can radically reduce the amount of repetitive human work involved in large, semi-structured technology migrations.

Westpac still needed skilled engineers, governance, project management, and quality control.

AI changed the scale of what that team could accomplish.

For other enterprises, the most important takeaway is simple:

Track total economics, not just AI-token spend, and deploy AI where repetitive work is large, measurable, and easy to validate.

Sources Consulted

iTnews, Westpac spends US$12,000 in AI tokens on intranet migration, September 18, 2026.
Westpac, Westpac and Amp Frontier to partner on AI engineering, August 7, 2026.
Westpac Investor Centre, Data, Digital and AI Update, September 2026.
Westpac employee project update describing the MyW migration scale and 61-hour execution.
Westpac 2026 Interim Results presentation covering company-wide AI adoption and Copilot use.
Bloomberg Law, AI Costs Lead Westpac to Prod Staff Toward ‘Sensible’ Model Use, July 2026.

Editorial Transparency Note

Claims requiring final verification: Token spending, employee counts, project metrics, and AI architecture can change or be clarified by Westpac. Recheck any newly released Westpac statement before publishing.

Expert review: Not required for a general enterprise-technology news explainer, though technical review would improve any deeper claims about migration architecture or model routing.

Disclosure: Do not claim Aiera.blog independently audited Westpac’s cost savings or migration results unless such verification was performed.