That Won't Pass Audit

After nearly 20 years advising public and private companies, Ro Sokhi has seen just about all there is to see in the world of finance. His career has been defined by helping organisations manage growth, risk and the increasingly scrutinised regulations. Today, as a partner at UHY, Ro still works closely with companies in the midst of navigating high-stakes financial transitions. Even as the finance function is being reshaped by AI.

Over the past few years, Ro has watched AI fundamentally change how finance teams operate. With organisations facing “a genuine shortage of finance and accounting talent,” he says, attention has increasingly been turned to AI and “emerging agentic tools that promise to alleviate this labour crunch”. While this potential is clear, Ro notes that mass adoption hasn’t yet taken hold, leaving many teams struggling to navigate this change.

For many organisations, that transition has been made even tougher because finance ecosystems have become more complex. As it stands, most businesses haven’t seen solutions that can “reliably replace humans out of the loop,” as Ro describes, or reduce labour hours at a significant scale. That means they’re still “carrying largely the same staffing footprints to cover the gaps in AI capabilities”.

For finance leaders feeling this pressure, Ro’s advice is to take care. While there is “significant promise from AI and emerging agentic tools,” he notes that these platforms remain highly dependent on human review. As a result, finance leaders may benefit from waiting until there is a “clear visibility into how they will meaningfully reduce costs” before committing to major investments.

While Ro’s advice around caution is clear, it’s undeniable that these tools have already had a major impact in the finance space. With that in mind, we started our interview by asking just how much has already changed in the last few years.

How has the role of the finance function changed in the past three to five years, and what do you think most organisations still underestimate about that shift?

Over the past three to five years, the finance function has been reshaped by a combination of labour constraints, technology promises (particularly AI), and structural shifts in how work gets done. First, organisations are facing a genuine shortage of finance and accounting talent. University pipelines have shrunk, retirements have accelerated, and demand for finance professionals has surged, creating sustained pressure on finance teams that were already lean.

Second, there has been growing attention on AI and emerging agentic tools that promise to alleviate this labour crunch by automating core finance activities such as invoice matching, reconciliations, and financial reporting workflows. While these technologies are potentially transformative, mass adoption has not yet taken hold, and most organisations have not seen solutions that can reliably replace humans out of the loop or materially reduce labour hours at scale.

Third, many companies have responded by increasing reliance on outsourcing and third-party advisory firms. Particularly following the Covid-era labour shortages, outsourcing functions such as accounting, tax, accounts payable, and even fractional CFO support became more practical than competing for scarce talent in a low-unemployment environment.

Finally, increased automation around revenue and financial reporting has driven greater reliance on IT systems, resulting in heightened scrutiny of general IT controls by external auditors. This has been especially challenging for smaller or emerging technology companies, including those in fintech and crypto companies, that rely heavily on home-grown systems with less mature control environments.

Which accounting or finance processes are still far more manual than they should be, and what’s stopping teams from automating them?

Despite a clear surge in automation and technology implementation in the finance function over the past 20 to 30 years, various functions can still be very manual, particularly when not using large, purpose-built ERP systems.

Emerging growth companies are ironically less likely to use sophisticated accounting software (instead opting for simpler systems like QuickBooks), requiring manual account reconciliations, revenue workflows and accounts payable functions. Even with larger organisations, we regularly see finance teams exporting ERP data into spreadsheets to reconcile cash, intercompany, inventory or accruals, then emailing files back and forth for approvals.

Revenue can be particularly manual: contract reviews, performance obligations tracking and billing reconciliations often live in shared drives and inboxes. On the payables side, invoice coding, exception handling and duplicate detection can still be human-heavy even when an AP platform exists.

While there is significant promise from AI and emerging agentic tools to alleviate this, we have yet to see large-scale adoption of AI tools in the finance function. Most AI implementations today remain narrow, supervised and highly dependent on human review. As a result, many finance leaders find themselves in an awkward interim state, experimenting with AI-enabled efficiencies so as not to get left behind, but still carrying largely the same staffing footprints to cover the gaps in AI capabilities.

What are the biggest integration challenges between finance systems (ERP, payroll, billing, reporting), and how do they affect accuracy and agility?

Finance systems have historically been built in silos. Companies may end up with a separate general ledger system, payroll platform, billing platform, CRM and BI tool, each with its own customer IDs and workflow mappings. While larger integrated ERP systems like SAP and Oracle do exist and may integrate well with other enterprise-level platforms like Salesforce, these platforms are still the domain of larger organisations.  We find that smaller tech-enabled startups are unable and unwilling to invest the time and (considerable) money required to set up and maintain enterprise-level solutions.  

There are middle-market-friendly solutions from newer and smaller payroll and billing platforms with APIs, but picking the right platforms that function well together at the forefront can be a challenge. And integrations between smaller platforms could be costly, sometimes requiring implementation services billed at hourly rates. Because of these challenges, we find that most of our smaller tech startup clients use platforms that offer a “best-fit-for-most,” and manually reconcile or import anything else that may not have easily or economically integrated well (eg payroll). 

What skills do modern accountants and finance professionals need today that weren’t essential five or ten years ago?

As every year passes, the finance function is expected to become more and more IT-literate. This was true ten years ago, and this is true today. This can include being able to work with larger and larger data sets, including performing data analysis, summarisation, basic querying and data visualisation for presentation to stakeholders and investors. Being able to communicate a cohesive story from the data is key; leaders need data that can explain drivers of what changed, why it changed, and what should be done next. 

Additionally, there is an emerging need for AI and governance literacy, understanding where automation helps, where there are opportunities and where there are still gaps in the technology, where it introduces risk, and how to put guardrails in place. 

What’s one finance or accounting technology investment that delivered unexpected value, and why?

Tech startups need to be frugal with every purchase, so it is understandable that accounting and finance tech stacks are at the bottom of the priority list. We routinely see our smaller clients select general ledger platforms like QuickBooks. Although these platforms may appear to be cost-effective upfront, they may be more costly on the backend, as later explained. 

We have found that clients who invest early in implementing middle-market-friendly ERP platforms like NetSuite and Sage benefit later on. These platforms make it easier to get through audits and protect sensitive information. This is because these platforms allow the system to be “locked down” for control purposes, offering the ability to dial-in read/write access, the ability to appropriately segregate functions (eg marketing team should not have access to revenue reporting function), and the ability to achieve effective general IT controls (such as change management). 

In fact, inexpensive platforms without these capabilities may be more expensive in the long run, requiring more manual controls and interventions to compensate for the lack of an effective internal control environment, including manually reviewing posted journal entries and performing detailed account reconciliations and variance analyses.

Having an effective control environment allows for faster and cheaper audits and sell-side due diligence work. This allows for faster access to capital, lower costs during M&A transactions, credibility with sophisticated stakeholders, ability to scale without disruption, and reduced operational risk.

What advice would you give to finance leaders who are under pressure to ‘digitize’ but lack internal buy-in or technical resources?

Start by reframing digitisation as an opportunity for growth rather than merely a risk and compliance matter. If stakeholders only hear “new system,” they assume disruption and cost. Instead, quantify pain in business terms: inability to raise capital, spiralling rework hours, excessive audit findings and cost overruns, billing leakage, slow quoting, and even missed forecasts. 

Second, choosing the right ERP platform will pay dividends down the road. Companies that invest earlier in middle-market ERPs like NetSuite, Sage or the like benefit from stronger segregation of duties and general IT controls, which materially reduce audit effort and operational risk. The result is faster, cheaper audits and diligence, improved credibility with investors, and greater flexibility to raise capital or scale without disruption.

Finally, while there is significant promise from AI and emerging agentic tools, these platforms remain narrowly focused and highly dependent on human review. As a result, finance leaders may benefit from being cautious in expending too much capital on these platforms until there is clear visibility into how they will meaningfully reduce costs in other areas of the organisation.