Artificial intelligence has moved from theoretical curiosity to a practical tool faster than most of us expected. In valuation work, AI can now (theoretically) analyse financial statements in seconds, spot trends across datasets, generate forecasts, and benchmark performance against thousands of comparable businesses.

It's impressive. It's also incomplete.

The question isn't whether AI is useful in valuation, because it clearly is. The question is where the boundaries sit, and what happens when those boundaries get ignored.

AI excels at tasks that involve pattern recognition, speed, and scale. In valuation, that translates into some genuinely helpful capabilities:

• Rapid financial analysis. AI can process years of accounts, highlight anomalies, and flag inconsistencies far faster than manual review. It's particularly good at spotting outliers such as unexpected drops in margin, sudden shifts in working capital, or trends that don't align with industry norms.

• Benchmarking at scale. Where traditional valuation might compare a business against a handful of similar companies, AI can pull data from thousands of transactions, sector reports, and market databases. That breadth can surface insights that would otherwise take weeks to uncover.

• Document summarisation. Whether you’re dealing with legal agreements, management accounts, or board minutes, AI can extract key information and summarise dense documents quickly. That saves time and helps ensure nothing critical gets missed.

These are real efficiencies. They make valuation work faster and, in some cases, more rigorous. But they don't replace the work itself.

The limitations of AI in valuation aren't minor technical issues. They're fundamental to how AI works, and they matter most when the stakes are highest.

• The non-deterministic problem
  
  Most modern AI systems, particularly large language models, are non-deterministic. That's a technical way of saying that the same input can produce different outputs on different days.
  
  This isn't a bug, it’s a feature. AI models rely on probabilistic processes, not fixed logic. And for valuation work, that's a serious problem.
  
  A valuation needs to be defensible. It needs to be reproducible. If you're preparing a report, you can't have a model that produces one conclusion on Monday and a different one on Wednesday with identical inputs.
  
  Shareholders expect consistency. They expect transparency. They expect someone who can explain, line by line, how a conclusion was reached and be cross-examined on it. AI can't do that. It can't even guarantee it will give you the same answer twice.

• Confidentiality and data exposure
  
  When you feed information into most AI platforms, that data passes through external servers. Depending on the platform, it may be stored, logged, or used to train future models. For client financial data, especially in contentious or sensitive situations, that's an unacceptable risk.
  
  Shareholders don't expect their accounts, forecasts, or strategic plans to be uploaded to third-party systems. Advisors have a responsibility to protect that information, and using AI without proper safeguards can undermine that trust.

• Context that doesn't fit a dataset
  
  Business value isn't just a function of revenue and profit. It's shaped by relationships, leadership dynamics, customer concentration, supplier dependencies, and operational resilience. These are human factors, and AI struggles with them.
  
  A business might look healthy on paper, but be entirely dependent on one founder who's planning to retire. Or it might have strong financials but a toxic culture that's driving key people out. Or it might have contracts that look solid but relationships with clients that are fragile.
  
  AI can't assess these things reliably. It can't interview a management team, sense tension in a boardroom, or evaluate whether a business will function without its current leadership. But these are often the factors that matter most.

None of this means AI should be avoided. It just means it needs to be used appropriately.

AI is excellent for sanity checks and reasonability testing. Once a valuation has been prepared, AI can quickly flag whether assumptions fall outside normal ranges, whether multiples seem inconsistent with sector norms, or whether forecasts diverge from historical performance in ways that need explaining.

It's also useful for initial analysis, such as pulling together data, identifying trends, and structuring information so that human judgment can be applied more efficiently. In that sense, AI accelerates the process without replacing the professional doing the work.

The key is keeping AI in a supporting role, not handing it the conclusion.

At bizval, we use AI where it improves speed and rigour, but we don't rely on it for judgment or defensibility.

We've built proprietary models that sit outside generic AI platforms. That means client data stays protected. It's not uploaded to external servers, and it's not used to train third-party systems. Confidentiality isn't negotiable, and our approach reflects that.

AI helps us with analysis, benchmarking, and efficiency. It doesn't make the valuation decisions. Those are made by professionals who understand the business, the context, and the legal or commercial purpose the valuation needs to serve. Because here's what matters in high-stakes situations: it's not just about getting an answer. It's about understanding the levers that drive value, being able to defend the conclusion under scrutiny, and providing a narrative that makes sense to the people who need to rely on it.

AI can't stand in the witness box. It can't explain why it chose a particular method or discount rate. It can't adjust its approach when new information emerges. And it can't take responsibility when things go wrong. A valuation professional can.

But perhaps the most important part of our approach isn't about the technology at all; it's about the conversation. Valuations don't happen in a vacuum. They happen during divorces, shareholder disputes, succession planning, and moments when founders are genuinely uncertain about what comes next. The shareholder sitting across from us isn't looking for jargon or technical complexity. They're looking for clarity, and they're looking for someone who understands what this actually means for them personally.

That's where the human element becomes non-negotiable. We engage with owners where they are, not where we think they should be. We take the time to understand their situation and not just their balance sheet.

The future of valuation isn't AI instead of humans. It's AI working alongside professionals who know when to trust the technology and when to override it. That's the balance that delivers clarity, defensibility, and confidence, especially when it matters most.

Engineering that works with Chris Hutchinson 

In this episode, Graham speaks with Chris Hutchinson, founder of the Trebuchet Group, about what happens when you apply engineering discipline to the messy reality of leading people. The conversation explores why technical expertise alone doesn't create great teams, how one leader can transform an entire organization overnight, and why most attempts at systems improvement fail before they start.

They discuss the trap of working in your business instead of on it, why asking "why" is often the wrong question, and how giving people permission to struggle might be the most important thing a leader can do. 

Listen Here.

Building Omne with Candice Mullins 

In this episode, Graham speaks with Candice Mullins, co-founder and managing director of Omne, about building an accounting and advisory practice from a small Eastern Cape town that serves clients across South Africa and internationally.

They discuss the challenges of adopting cloud technology years before it became standard, managing remote teams before COVID made it mainstream, and building a partnership with co-founder Simon White that's held through growth, regulation, and the constant pressure small practices face. This is a conversation about staying grounded while scaling, choosing lifestyle over maximum revenue, and building a business that works for the life you actually want, not the one you're supposed to want.

Listen Here.

bizval featured on the Dinis Guarda podcast

In this episode, Graham joins Dinis Guarda for a wide-ranging conversation about what business valuation really looks like when you strip away the hype, and focus on the decisions founders actually need to make.

They discuss the bizval vision, why “value” is often misunderstood (especially around IP), and how AI can support clearer thinking without replacing the human judgement that makes a valuation defensible. It’s a grounded look at valuation as a practical tool: for growth, exits, funding conversations, and getting honest about the reality of your business.

Listen Here.

Bringing humanity back into high finance with Yeshvika Bahadur

In this episode, Graham speaks with Yesh Bahadur about something the valuation world rarely addresses directly: the human weight behind every transaction.

Yesh's work sits at the intersection of finance and psychology, and her perspective challenges the idea that deals are purely rational events. She and Graham discuss how burnout shapes decision-making, how leadership trauma quietly erodes value, and why founders sometimes struggle to exit even when the numbers tell them it's time.

Listen Here.

We recently co-hosted a workshop with Nick Bradley (author, speaker, and business growth expert) about the operating frameworks private equity uses to build companies that command premium valuations.

Nick walked us through why most businesses are worth less than their owners expect, how buyers actually assess value, and what separates businesses that attract investment from those that don't.

We also discussed the end-game principle, why founder dependency suppresses multiples, and the specific systems that drive valuation upwards, whether or not you're planning to sell.

If you suspect that your business might be more dependent on you than it should be, or if you want to understand what "investor-grade" actually means in practice, then check out the replay below – there’s plenty of good insights to dig into!

Watch the replay here.

2025 has been a full year for bizval. It’s been busy, challenging in the right ways, and ultimately very rewarding.

We've worked with more clients across more sectors than ever before. We've supported advisors navigating complex transactions, helped families through difficult situations, and provided clarity when it mattered most. None of that happens without the trust our partners and clients place in us, and we don't take that lightly.

To our team: thank you for the care, rigour, and professionalism you bring to every engagement. To our partners and collaborators: thank you for continuing to work with us and for the referrals, feedback, and conversations that make us better. And to our clients: thank you for trusting us with work that genuinely matters.

As we move into 2026, we're looking forward to deepening relationships, refining how we work, and continuing to deliver valuations that stand up when they need to.

From all of us at bizval, we hope you have a restful festive period and a strong start to the new year.

Until next time, 
The bizval team