Stronger billing control. Hours of manual review removed.
How Mirror Financial Services used Claude to identify billing exceptions earlier, reduce timesheet administration and give directors clearer visibility across the practice.
The client
Mirror Financial Services is a Pretoria-based accounting, audit and tax practice serving individuals and businesses across South Africa.
As the practice grew, billing, timesheet reviews and payment allocation required increasing amounts of manual administration. Mirror wanted to reduce this workload while keeping its existing systems, professional controls and human approval processes in place.

The challenge
Mirror’s billing process relied on information spread across its practice management system, individual timesheets, Excel billing sheets, its invoice register, bank records and Pastel.
This created several challenges:
Billing and timesheet administration required more than 40 hours during a typical month.
Timesheets were exported separately and reviewed manually.
Directors had limited visibility into time recorded compared with fees billed.
Billing inconsistencies were sometimes discovered only after invoices had been issued.
Incomplete bank references made matching payments to invoices time-consuming.
Much of the billing workload was concentrated at month end.
Mirror already had the core systems it needed. The missing element was an intelligent review layer that could compare information across those systems and highlight issues before the team acted.
The solution
Let’s Learn AI began by mapping Mirror’s billing, timesheet and payment processes using the practice’s own working files.
We then designed and implemented four Claude Projects inside Mirror’s own Claude Team workspace.
Billing Reviewer
The Billing Reviewer checks the monthly billing sheet before invoices are raised. It identifies potential issues such as:
Unusual fee differences
Duplicate or repeated entries
Missing dates or amounts
Client-name inconsistencies
Negative amounts and credit entries
Recurring clients missing from the billing sheet
Billing Insights
Billing Insights turns Mirror’s historical invoice register into a practical monthly management view. It provides:
Month-on-month and year-on-year billing trends
Billing patterns by client, service and biller
Recurring fees that may have stopped
Clients whose billing activity has declined
A monthly billing worklist
An indicative billing forecast based on historical patterns
Timesheet Review
The Timesheet Review analyses recorded time by employee, client, work type and activity.
It helps Mirror compare time spent with fees raised and highlights incomplete narratives, duplicate entries and other potential capture-quality concerns.
Payment Matcher
The Payment Matcher compares incoming bank deposits with outstanding invoices and proposes likely allocations.
It can handle abbreviated references, partial payments and payments covering multiple invoices. Where more than one allocation is possible, the Project refers the decision to a Mirror team member.
The results
The implementation delivered measurable value from its first application to Mirror’s operational data.
More than 30 exceptions identified before invoicing
During the first monthly billing review, the Billing Reviewer identified more than 30 exceptions requiring attention.
These included client-name inconsistencies, material billing differences and recurring clients missing from the billing sheet. The team could investigate these items before completing the invoicing process.
At least eight hours saved every month
The Timesheet Review saves the team at least two hours of manual review each week, returning more than eight hours during a typical month.
It also gives the directors clearer visibility into where time is being spent and how recorded time compares with fees billed.
Clearer management information
Claude brings information from separate operational files into structured exception reports and management analysis.
This gives Mirror’s directors a clearer view of:
Billing activity and changes
Time spent by client and service
Potential billing omissions
Client recovery and profitability indicators
Payments requiring investigation
“We have been impressed by how quickly Claude has delivered practical value. Work that previously required hours of manual review from our directors can now be analysed far more efficiently, while the final decisions remain with our team. Based on these results, we plan to introduce Claude into more workflows across the practice.”— Jandré Pienaar, Director, Mirror Financial Services
Adoption across the practice
All four Phase 1 Claude Projects were deployed inside Mirror’s own Claude Team workspace and incorporated into the practice’s billing, timesheet and payment-allocation processes.
The Projects support the employees responsible for billing, timesheets and payment allocation, while the directors use the resulting exception reports and analysis to guide their review and approval.
Following the initial results, Mirror plans to identify further workflows where Claude can reduce manual administration, strengthen internal controls and give its directors better access to management information.
Responsible AI with human oversight
Claude assists with checking, comparison and analysis, while Mirror’s team remains responsible for every final decision.
No invoice, payment allocation, accounting entry or client communication is finalised without human review and approval.
Mirror’s information is handled according to agreed data rules. The Projects are designed to flag uncertainty and missing information rather than make unsupported assumptions.
Practical AI built around the business
The implementation allowed Mirror to introduce AI without replacing its existing practice management and accounting systems.
Claude provides an additional control and analysis layer across the information those systems already produce. The result is less manual review, earlier identification of potential billing issues and clearer visibility for the practice’s directors.
Results at a glance
30+ exceptions | 8+ hours saved monthly |
Identified for review during the first monthly billing review | Based on at least two hours saved on timesheet review each week |
Could your business be losing hours to repetitive checking and administration?
Let’s Learn AI helps South African teams identify practical AI opportunities, build responsible workflows around their existing processes and train employees to use Claude effectively.

Comments