AI Revolutionises Financial Forecasting: Balancing Automation with Human Insight
Forecasting has long been the cornerstone of strategic financial management. At its best, a robust forecast guides investment decisions, cash management, and executive planning. But in too many businesses, it's still a manual, time-consuming exercise rooted in spreadsheets and institutional memory. Even for fast-growing businesses with cloud accounting and BI tools in place, forecasting remains a pain point.
Now, artificial intelligence is changing that. Quietly but steadily, AI is beginning to take on parts of the forecasting process that were once the domain of analysts and CFOs. And while the market is awash with hype, there are real, tangible shifts happening underneath. According to PwC's 2025 AI Business Predictions, AI adoption is accelerating across industries, with finance leaders prioritising tools that enhance predictive analytics and decision-making speed.
Similarly, Google Cloud's report on 2025 AI trends for financial services highlights how AI is enabling more accurate forecasting through advanced data processing and scenario modelling.
This article explores the current state of AI in forecasting, where it genuinely adds value, where human judgment remains irreplaceable, and what finance leaders need to know before automating too far, too fast. Drawing on recent research and real-world examples, we'll position CFOPartners as your guide in navigating this evolving landscape.
The State of Forecasting Today: Manual, Siloed, Stale
Many finance teams still rely on a combination of:
Exported data from accounting platforms (Xero, MYOB, QuickBooks)
Google Sheets or Excel models
Ad hoc assumptions, often stored in a leader's head
Even in businesses with $5M–$50M in revenue, we regularly see forecasts that:
Break when copied into a new tab
Are built once per quarter, if that
Take 1–2 weeks to complete
Don’t reconcile to actuals until month-end is closed
Recent statistics underscore this reliance on outdated methods. A 2025 AFP report reveals that 100% of FP&A professionals use spreadsheets for planning and reporting at least quarterly, despite growing frustrations with their limitations.
Additionally, 45% of respondents in a Fathom survey noted that spreadsheets become unmanageable as data complexity increases, leading to errors and inefficiencies.
Accenture predicts that by 2025, automation will handle 65% of repetitive financial modeling tasks, signaling a shift away from these manual processes.
The result? Forecasting becomes a backward-looking compliance task instead of a forward-looking decision tool. In 2025, with economic volatility persisting—rising interest rates and supply chain disruptions—this stasis is increasingly untenable.
Where AI Is Already Creating Real Value
The best applications of AI in forecasting today are focused not on replacing the CFO, but on accelerating the grunt work and increasing decision velocity. As highlighted in BCG's 2024 report on dynamic steering in financial planning, AI technologies like machine learning can make processes far faster and more accurate by analyzing vast datasets in real-time.
Here's where AI shines:
1. Data Structuring and Categorisation
AI tools can now ingest messy exports from accounting systems or bank feeds and:
Auto-categorise GL lines
Detect and surface anomalies
Allocate costs based on usage, seasonality, or driver-based logic
This alone can save hours per month in model prep and variance analysis. For example, tools like DataRobot enable adaptive, high-precision forecasting by learning from real-time data, reducing cash flow uncertainties by up to 30% in some cases.
2. Narrative Explanation and Reporting
Using large language models (LLMs), AI can now:
Summarise forecast changes in plain language
Generate board-level commentary on cash runway, margin shifts, or sales projections
Translate complex movements into something understandable for non-finance stakeholders
This has immediate application in board packs, investor updates, and management reporting. Generative AI, as noted in NVIDIA's 2025 State of AI in Financial Services, is being used for faster, more accurate financial narratives, enhancing communication across teams.
3. Scenario Modelling at Scale
AI can now model dozens of what-if scenarios without the user needing to rebuild tabs. For instance:
What if conversion rates fall 10%?
What if supplier costs increase 12% in Q4?
What if we hire 5 more salespeople next month?
These scenarios can be generated, ranked by financial impact, and surfaced automatically. Jedox's AI-driven tools, for example, improve scenario planning efficiency, allowing finance teams to evaluate multiple outcomes in minutes rather than days.
Research from Cube Software shows that AI forecasting boosts accuracy by 20-40% through dynamic model adaptations.
Where AI Still Falls Short
While AI can speed up mechanical tasks, it lacks the context, judgement, and strategic framing that an experienced finance leader brings. A 2025 study on AI in financial forecasting highlights persistent limitations like model opacity and data quality concerns.
1. Understanding Strategic Trade-Offs
AI can suggest cuts or investments, but it can’t weigh:
Short-term cash preservation vs long-term growth
Hiring vs automation in a shifting talent market
Expansion risk vs shareholder pressure
These decisions are fundamentally human. As Risk.net points out, AI cannot consistently predict market movements due to unpredictable human behaviours and external shocks.
2. Judging the Trustworthiness of Inputs
AI will forecast based on the inputs it’s given. If your Shopify feed has returns misclassified, or payroll timing is off, the model won’t know. A human can spot that. AI can't (yet). Cube Software notes that AI requires vast amounts of accurate data to avoid biases and inaccuracies.
3. Explaining Confidence (or Lack Thereof)
AI can generate a forecast. But when a board asks, "How confident are we in this number?" — only a person familiar with the data, team, and commercial realities can give a meaningful answer. Ethical implications, including bias and lack of transparency, further complicate reliance on AI alone, as explored in a ResearchGate paper.
The Future Forecasting Stack: Human + AI
We believe the future isn't AI or human judgment. It's both, layered intelligently. Workday's 2025 insights on AI in corporate finance emphasise this hybrid approach, where AI streamlines processes while humans provide oversight.
The new forecasting workflow looks like this:
AI ingests source data and proposes a baseline forecast
AI highlights anomalies, risks, and sensitivity ranges
Finance leaders sense-check and adjust based on strategic insight
Narrative commentary is generated and reviewed
Outputs are fed directly into board packs or dashboards
This model is already being trialled in forward-thinking businesses. It changes the role of finance from "builder of models" to "interpreter of signals." Agentic AI, projected to drive a $244 billion market in 2025 per Statista, is powering this next generation of FP&A.
What Finance Leaders Need to Watch For
If you're a founder, CFO, or strategic advisor, here are the considerations before you implement AI-led forecasting:
Source Integrity: AI is only as good as your inputs. Fix your chart of accounts, bank rules, and CRM before expecting automation to save you. PreferredCFO warns that improper AI use can lead to errors if data foundations are weak.
Explainability: If you can't explain how the forecast was built, you can't defend it. Tools that allow override and traceability matter.
Governance: Forecasts inform hiring, cash runway, and strategy. They are not a playground for unvalidated AI outputs. Infosys BPM stresses maintaining benchmarks for anomalies like market crashes.
Cost of Change: AI systems require new workflows and sometimes team retraining. Factor that into ROI.
Why This Matters Now
With cost pressure rising and capital tighter than it’s been in a decade, businesses need:
Faster insights
More agile reforecasting
Clearer communication of risks and trade-offs
AI can help with all three, but only if implemented with strategic oversight. Citizens Bank's 2025 AI Trends report notes increasing budgets for AI in finance, driven by these pressures.
This is where finance leaders matter more than ever. Not because they need to write every formula themselves, but because they guide what matters, what needs context, and what decisions flow next.
CFOPartners' Role
At CFOPartners, we help growing businesses build forecasting systems that balance automation with insight. Whether it’s designing a 3-way model for investor readiness, or embedding AI tools into existing processes, we act as the strategic layer between systems and decision-makers.
In the months ahead, we’ll be sharing more insights on:
The role of forecasting in exits and valuations
Benchmarks on forecast quality in mid-market businesses
Real-world AI use cases (and failures)
If you’re rethinking your forecasting process, or want to understand where AI can actually make a difference reach out. We’d be happy to have the conversation.
hello@cfopartners.com.au
