Pengana, a firm in the financial services industry, has moved an AI-powered client service idea from concept into production. Built with First Focus, its Outlook-based AI assistant helps employees draft responses to common client enquiries faster, while sensitive information is redacted and people retain control over every response.
The opportunity is bigger than faster email writing. Pengana is testing whether AI can increase service capacity during enquiry spikes, reduce repetitive work and improve client responsiveness without compromising governance or personal service.
Hear directly from Pengana about how the idea moved into production, the unexpected challenges the team encountered, how sensitive information is protected and how the business plans to measure the value created.
Pengana’s first implementation provides a practical example of what an AI-enabled client service workflow can deliver:
Faster client responses. Employees can start with a relevant draft instead of a blank email, reducing the time required to respond to common enquiries.
Greater capacity during demand spikes. AI can help absorb repetitive enquiries around tax periods, market events and capital raises when client demand rises quickly.
Less repetitive work. Employees can spend more time on complex, sensitive or highly personalised client matters instead of repeatedly drafting similar responses.
More consistent responses. The assistant can draw on approved information when addressing recurring questions, helping the team maintain consistency at scale.
Human control remains. Employees review, personalise and approve communications before they are sent, keeping judgement and accountability with the Pengana team.
Privacy controls are built into the workflow. Personal and sensitive information is redacted before the relevant content reaches the AI.
A measurable path to ROI. Pengana is tracking utilisation and plans to compare AI-assisted activity with previous manual workloads to estimate the staff capacity released.
We’re being more responsive. We can actually start servicing a lot more clients.
One of the most useful parts of Pengana’s experience is that the biggest implementation questions were not about whether AI could write an email.
They were operational.
How do you prevent sensitive client information from reaching the AI? What happens when several employees use the same shared mailbox? How do you avoid duplicate processing? Which enquiries should remain entirely human-managed? And how do you know whether the system is creating enough value to justify expanding it?
Those questions shaped the solution that eventually went into production.
And they are also what makes the Pengana story particularly useful for other business leaders.
For Pengana, the starting point was not AI.
It was client service.
In financial services, bringing a client into the business is only the beginning of the relationship. Clients continue to need information as markets change, government announcements are made, investments evolve and tax obligations arise.
As Varun explains:
You actually have to service the clients that you’ve brought in.
Some of those enquiries are complex and highly individual.
Many others are variations of questions Pengana has answered before.
When a large number arrive at once, employees can spend significant time repeatedly locating approved information and rewriting similar responses.
That creates an opportunity for AI.
Instead of starting each response from scratch, the assistant can interpret the enquiry and generate a relevant draft using information available to it.
The employee then reviews the response, makes any necessary changes and decides what ultimately goes back to the client.
The business outcome is not simply faster writing.
It is more capacity to respond when clients need answers.
The value of the workflow becomes particularly clear when enquiry volumes are not evenly distributed.
Pengana identified several situations where many clients may ask similar questions within a relatively short period:
Tax statements are a simple example.
A large number of clients may want to know when their statement will be available because they need to pass that information to an accountant or tax agent.
The underlying answer may be relatively straightforward.
But if the same question arrives dozens or hundreds of times, manually drafting each response consumes capacity that could be spent elsewhere.
AI changes the economics of that repetitive work.
It allows the team to handle more enquiries without treating every email as an entirely new drafting exercise.
For business leaders considering AI, this suggests an important way to evaluate potential use cases:
Don’t only look for work that happens frequently. Look for work where demand can suddenly exceed the team’s normal capacity.
That is where automation can have a disproportionate impact.
One of the most distinctive lessons from the project came from something that initially sounded simple.
Pengana’s client inbox is shared by multiple employees.
Once automation is introduced, that creates a new question: what happens if the AI workflow is triggered simultaneously across several accounts?
Without the right design, the same enquiry could potentially be processed more than once.
That creates risks such as:
The team went through several iterations to determine how the assistant should operate safely inside the shared mailbox.
This is a useful reminder that successful AI implementation is not only about choosing an AI model.
The surrounding workflow can be just as important as the intelligence inside it.
A model might be perfectly capable of generating a useful answer while the overall solution still fails because the process around it was poorly designed.
That type of operational detail only becomes obvious when AI moves from a demonstration into a real business environment.
Data governance was considered before development began.
That matters in a financial services environment where client emails can contain names, identification documents, banking information and other sensitive personal data.
Pengana did not want that information passed unnecessarily into the AI workflow.
A redaction stage was therefore designed into the process.
Information identified as personal or sensitive is filtered before the relevant content is sent to the AI. The assistant receives what it needs to understand the underlying enquiry, rather than the full set of personal information that may have appeared in the original email.
The generated response is also designed to avoid automatically inserting sensitive client-specific information.
The principle is simple:
Define what information the AI is allowed to see first. Then build the AI solution inside those boundaries.
For businesses operating in regulated or information-sensitive environments, that may be a much more useful starting point than beginning with model selection.
AI drafts the response. People make the decision.
Pengana deliberately retained human oversight.
We still have oversight on what goes out.
The AI creates a draft rather than autonomously responding to the client.
That distinction matters.
Some questions can be answered using approved, commonly used information.
Others involve:
Employees retain the discretion to ignore the AI workflow and respond manually when that is more appropriate.
The objective is therefore not to remove people from client service.
It is to allow technology to handle part of the repetitive drafting work so employees can spend more of their time where human expertise adds greater value.
At a high level, Pengana and First Focus designed the process around several stages:
Mapping this flow proved to be an important part of the project.
Varun explains that the initial idea evolved considerably once the team began working through exactly how the solution would need to operate.
There were decisions around redaction, model selection, shared mailbox behaviour, security, workflow timing and other practical considerations.
Once the process itself had been mapped and the major edge cases addressed, the build progressed relatively quickly.
The lesson for other organisations is significant:
A successful AI project may require more process design than prompt engineering.
Varun did not begin the AI Investment Fund process with every technical detail already decided.
He understood the problem and had an idea for how AI might help.
The architecture developed through collaboration with First Focus.
That included evaluating different technical options and identifying issues that were easy to overlook when thinking about the idea at a high level.
The shared mailbox was one example.
Varun’s initial assumption was that an incoming enquiry could simply trigger the AI process.
The implementation work revealed another question: what happens when multiple accounts are interacting with that same process?
That type of consideration is what separates an interesting AI concept from a reliable business workflow.
It also illustrates where external expertise can create value.
The person closest to the business process may understand the problem extremely well without necessarily knowing every architectural, security or integration consideration required to put the solution into production.
Combining those two forms of expertise can be powerful.
One of the strongest parts of Pengana’s approach is that the team is not relying only on anecdotal feedback to determine whether the project worked.
When the AI workflow is used, a specific tag is applied.
That gives Pengana a way to measure utilisation.
Varun plans to compare the current workflow with a previous period in which similar enquiries were handled manually, particularly around a previous capital raise.
The team can examine:
A simplified way to think about the measurement is:
AI-assisted enquiries × average manual handling time = indicative staff capacity released
Pengana had not established a final financial ROI figure at the time of the conversation.
That distinction is important.
The project has moved into production, but the team is still gathering the evidence required to quantify its impact.
What Pengana does have is a practical method for measuring that impact rather than declaring success simply because an AI tool has been deployed.
There was some scepticism when the assistant was first introduced.
Pengana did not try to solve that by forcing every enquiry through AI.
The manual option remained available.
Employees could use the draft when it helped and respond themselves when it did not.
Over time, usage increased as the team began recognising repetitive enquiries where the assistant could remove much of the initial drafting work.
That points to another practical lesson for AI adoption:
A useful AI system can earn adoption by removing friction from an existing workflow.
Employees do not necessarily need to be convinced by an abstract AI strategy if the benefit appears directly inside work they already perform.
Pengana is now putting the deployed assistant to work around a capital raise.
That provides a useful test environment.
Capital raises can create repeated enquiries from prospective investors seeking information about the opportunity.
Many of those questions can relate to material already contained in approved FAQs or supporting information.
The assistant can help surface that information in response to the specific question being asked, rather than requiring employees to repeatedly construct the same answers or expecting prospective investors to search for the information themselves.
Human oversight remains important, particularly where eligibility, personal circumstances or sensitive information are involved.
But this moves the project beyond controlled testing.
Pengana can now observe how the workflow performs during the type of real-world demand spike it was designed to address.
Pengana is already considering where the same pattern could create value elsewhere.
We’ve barely scratched the surface.
Sales is one potential area.
Many organisations have experienced employees who effectively become internal “brain banks”. Because they hold significant knowledge, people repeatedly approach them with questions.
The employee may know the answer immediately.
The challenge is that their capacity remains finite.
The same pattern can appear across:
If a knowledgeable employee repeatedly turns existing information into answers for other people, there may be an opportunity for AI to assist with retrieval and drafting.
Pengana’s first deployment therefore creates value beyond the emails it helps draft today.
It gives the organisation a working example it can use to identify similar patterns elsewhere.
Another important part of Pengana’s story is where the idea came from.
It came from someone close to the operational problem.
Varun had experienced the workload himself and could see the repetitive process that needed improving.
Pengana provided an environment where the idea could be proposed, supported and developed.
If you have an idea, pitch it. If enough people back you, you can roll with it.
For leaders building an AI strategy, that suggests a useful change in the questions they ask employees.
Instead of only asking:
“Where could we use AI?”
Ask:
“What work are you repeatedly doing that you already know how to do, but cannot do fast enough at scale?”
Or:
“Where does demand regularly exceed your team’s capacity?”
Or:
“Which questions do you find yourself answering again and again?”
The answers may reveal AI opportunities that are difficult to identify from the executive level alone.
Pengana began with a client-service problem, not a requirement to deploy AI.
The technology followed the use case.
High-volume periods can make a stronger business case for AI than average workload alone suggests.
Determine what the AI should and should not receive before building the workflow.
The shared mailbox issue shows how seemingly small operational details can determine whether an AI solution works reliably.
Drafting can be automated without automating responsibility.
Pengana’s tagging and historical comparison approach creates a clearer path towards demonstrating actual value.
Operational teams often have the clearest view of repetitive processes, capacity constraints and service bottlenecks.
Pengana’s experience provides a practical model for moving AI beyond experimentation.
Find a repeatable problem. Define the desired business outcome. Establish the information boundaries. Map the workflow and its failure points. Keep people in control where judgement matters. Then measure the new process against the old one.
The technology matters.
But Pengana’s experience suggests that much of the value comes from what surrounds it: process design, governance, employee adoption and a clear understanding of the business problem being solved.
That is what turns an AI demo into an operational capability.
Pengana’s experience also highlights why implementing AI successfully requires more than access to an AI model.
The solution needed data governance. It needed workflow design. It needed security controls. It needed to fit into the technology employees were already using. And once deployed, it needed adoption, measurement and ongoing improvement.
That is the thinking behind CORE, First Focus’s managed AI & IT service.
CORE brings AI, automation, data governance, cybersecurity, training and day-to-day IT management together as one ongoing capability. Rather than treating AI as a standalone project, the goal is to continuously identify where technology can remove friction, improve productivity and reduce risk across the business.
The Pengana project provides a practical example of why those pieces need to work together.
The AI model can generate the draft, but the business outcome depends on everything around it:
For many businesses, that is the bigger opportunity.
The goal is not to complete one AI project.
It is to build the capability to continually find processes that can be improved, implement the right technology securely and turn those improvements into measurable business outcomes.
Traditional managed IT has largely focused on keeping technology operational: supporting users, maintaining systems, resolving incidents and protecting infrastructure.
Those things remain essential.
But businesses now need their technology partner to do more than keep the lights on.
They need help answering questions such as:
That is where First Focus is evolving the managed service model through CORE.
As a managed AI & IT provider, First Focus combines the operational foundations businesses already depend on with AI, automation, data governance, security and ongoing capability building.
The objective is straightforward: less manual work, faster decisions, reduced risk and more capacity for growth.
Pengana started with a specific operational frustration: too much time being spent repeatedly responding to similar client enquiries.
Your opportunity may be somewhere completely different.
It could be buried in finance approvals, customer service, reporting, document management, sales administration, onboarding or an internal process your team has simply accepted as “the way we’ve always done it”.
CORE helps identify those opportunities and turn them into secure, managed improvements as part of your ongoing IT partnership.
Pengana built redaction into the workflow so personal or sensitive information can be removed before the relevant enquiry reaches the AI. Employees also retain responsibility for reviewing and sending the final communication.
Pengana’s experience suggests repetitive enquiries and periods of unusually high demand are particularly strong candidates. Examples include tax periods, market events and capital raises.
Yes, but Pengana’s implementation highlights the need to design carefully for multiple users. Without appropriate workflow logic, shared mailboxes can create risks such as duplicate processing.
No. The solution generates drafts. Pengana employees retain oversight and decide what is ultimately sent.
The workflow applies a tag when AI is used. Pengana can compare utilisation and estimated handling time with previous periods where similar enquiries were answered manually. The final ROI was still being evaluated at the time of recording.
Employees retained the ability to respond manually. Adoption increased as they encountered repetitive questions and experienced the assistant reducing the drafting work required.
The conversation identifies potential applications across sales, adviser relationships, institutional teams and other areas where knowledgeable employees repeatedly answer similar enquiries.
Pengana was one of five businesses selected through First Focus’s $100,000 AUD AI Investment Fund, with each successful business receiving $20,000 towards turning an AI-powered idea into a working solution.
This follow-up conversation explores what happened after the original idea was selected, including how Pengana approached governance, workflow design, adoption and measurement once the solution moved into production.
IT Consulting
Data Services
IT Infrastructure Projects
IT Procurement
Business Continuity Planning
Cloud Migration Services
Managed IT Services
IT Support Services
Managed Cloud Services
Managed SharePoint Services
End User Computing
Unified Communications
Connectivity
Virtual CIO
Offshore Staff Outsourcing (BPO)
Managed Security Services
ACSC Essential 8
Cybersecurity Risk Assessment
Cloud Backup and Disaster Recovery
Cyber Security Connect and Protect Program
Automation
IT Strategy
Microsoft 365 Consulting
Power BI
Careers
Awards and Certifications
Management Team
Case Studies
Sell Your IT Business
Our Commitment to the Environment
CSAT Charity Donations
Media Kit
Sydney
