MSI member firms demonstrate AI’s potential to transform audit

AI is reshaping how professional services firms deliver – not as a future possibility, but as a present reality.

In audit, it is enabling firms to prepare and process data more efficiently, analyse larger and more complex datasets, and, in selected procedures, test complete transaction populations rather than samples.

Within the MSI network, this is already happening. Over the past year, Cecil Kilpin & Co (Cape Town) and Aitken Lambert Elsworth Inc (Durban), both MSI member firms, collaborated with RAiN Chartered Accountants (South Africa) on live client engagements, applying RAiN’s proprietary AI platform, Monsoon, to work that had reached the practical limits of manual effort. RAiN’s team designed and delivered the AI-powered components while each firm retained full ownership of the client relationship.

Monsoon is ISO 27001:2022 (information security) and ISO 42001 (AI management systems) certified, ensuring data security and responsible AI governance are built into every engagement.

Use case 1 — Cecil Kilpin & Co: Reconciliation support for a commodities trading client

Challenge
Cecil Kilpin & Co was auditing a large agricultural commodities trading and animal-feed client. RAiN worked from a raw client data room of more than 1,500 files (over 700 spreadsheets, 550 PDFs, and underlying email correspondence) holding in excess of 1.5 million lines of transaction data across fourteen financing facilities.

The difficulty was structural: the client, its financing bank, and the mills that bought the grain shared no common transaction reference. The client invoiced the mills on the bank’s behalf for financing received through the entity, acting as agent, with only net proceeds remitted. The mills settled against weekly schedules – frequently before the bank had even posted the invoices they were paying. No two ledgers could be matched line for line. Despite significant effort from both the audit team and the client’s internal teams, the reconciling differences could not be isolated.

Approach
Cecil Kilpin & Co and RAiN collaborated, applying Monsoon to the engagement. RAiN’s team rebuilt the settlement logic from the client’s own weekly schedules, restoring a traceable path from every mill payment to the invoice, customer-ledger entry, and general-ledger posting behind it. Monsoon independently re-performed twelve months of daily value-dated interest and per-tonne fee calculations across all fourteen financing facilities.

Outcome
Every payment in the population now reconciles to the cent. Each remaining estimate is labelled on the row it affects, and items that genuinely require third-party documentation are listed explicitly rather than buried in a total.

Reconciling items that could not previously be pinpointed were identified and traced back to their origin. The audit firm got the evidence they needed. The client received clarity on reconciling items in their own records that had remained unresolved despite significant internal effort.

Business Impact
Cecil Kilpin & Co was able to complete the engagement with confidence. The client received assurance that the figures had been properly reconciled and the differences understood – an outcome that had not been achievable through manual effort alone.

Key Takeaway
This wasn’t about doing the same work faster. It was about solving a problem that manual effort could not solve. The data was there – but no human team could practically work across all of it at this volume and complexity. Monsoon made that possible.

Use case 2 — Aitken Lambert Elsworth Inc: Substantive income testing for a specialist client

Challenge
ALE needed to perform substantive testing of income for a client in the South African horseracing industry, covering a full year of transactions across dozens of individuals and multiple venues. Data from the client’s general ledger had to be reconciled to an independently sourced external dataset. The data needed for independent verification sat in non-standard sources outside the client’s own systems, making access and extraction a challenge in itself.

Under a manual approach, the firm would have tested a sample – not because sampling was the ideal methodology, but because full-population coverage was impractical within the time and resources available.

Sampling risk is a limitation auditors have always accepted.

Approach
ALE and RAiN collaborated, with RAiN’s team designing and executing an approach through Monsoon that tested every transaction against the independent source data – covering 2,834 individual races, recalculating the winnings and commission splits for each jockey and horse combination based on the actual race results – confirming that nothing was missing from the records and that every transaction that was recorded matched correctly back to its source.

All findings were fully traceable, allowing the auditor to re-perform and verify any finding independently.

Outcome
100% of transactions were tested. The engagement achieved complete population coverage where a manual approach would have covered only a sample.

Business Impact
ALE was able to offer their client assurance that every transaction in their records was verified –not a statistical estimate based on a sample, but complete coverage. A level of assurance that the previous delivery model could not practically provide.

Key Takeaway
Full-population testing is now genuinely achievable where manual approaches made it impractical. The result is the elimination of sampling risk – a limitation the profession has always lived with. The trade-off is a more intensive review process for the auditor, but the outcome is a level of assurance that sampling, by design, cannot match.

Why this matters
Both collaborations demonstrate the same principle: AI doesn’t just improve existing processes – it makes outcomes possible that weren’t achievable before. Whether it’s finding differences that no one could locate, or giving clients complete assurance across their entire dataset, the constraint was never capability or skill. It was the practical limit of what manual effort could cover.

RAiN’s approach is human-led throughout. The team designs the approach for each engagement, engineers the workflows, builds the context the AI needs to operate accurately, and validates every output. The AI executes. The professionals ensure it executes the right thing, the right way.

These collaborations between MSI member firms show what becomes possible when expertise, technology, and the trust of the MSI network come together.