Welcome to Issue 1!
Today's Topic: Practical thinking on AI for CEOs and finance leaders.
Welcome! Aperna is a fractional CFO firm working with small and mid-sized companies. This newsletter is for CEOs and finance leaders who want a finance perspective on AI.
In this issue, Brian Hendry, Principal of Aperna, covers what has changed with AI in the past six months, why it matters for your business, and what finance leaders should do about it. You should walk away with something you can use — a question to ask your team, a framework to apply, a mistake to avoid. We don't do "Which LLM is best", vendor reviews, or "top 10" lists — there are plenty of newsletters that cover these.
GETTING STARTED
This isn't AI hype — the shift is already happening
"If you're not using AI to move faster or make smarter decisions, you're behind. AI isn't just a tool, it's leverage."
What used to take weeks to code now takes minutes. Tasks that required teams are being handled by software running overnight. The question for CEOs and finance leaders isn't whether to engage with AI — it's whether you're building the judgment to use it well, and to protect your business from the risks of using it badly.
2026 is different — and here's why it matters
What has changed in the past six months is a combination of improvements that together cross a meaningful threshold:
Models are meaningfully more capable and consistent
They can execute repeatable multi-step actions reliably
Persistent memory across sessions is now achievable
AI agents can take real actions in business workflows
MCP (Model Context Protocol) enables AI tools connect directly to the applications your business already runs
These aren't incremental upgrades. Together they represent a shift from AI as a novelty to AI as infrastructure.
My own experience reflects both how far the tools have come and how far they still need to go:
A year ago, a venture capital firm published an AI-generated financial valuation model. I identified that it consistently used incorrect data pulled from SEC 10-K filings — errors subtle enough to pass a casual read but material enough to invalidate the analysis.
Around the same time, like many finance professionals, I found Microsoft Copilot in Excel unreliable for formula generation and financial analysis.
Two months ago, I flagged a similar data error in a financial model published by a different VC firm.
Last week, I asked Claude to write a Python script for data extraction. It generated an incorrect script that produced errors in the final report.
The pattern: AI tools can produce outputs that look authoritative and contain significant errors. The lesson isn't to avoid AI — it's that AI requires informed oversight, and finance leaders are well-placed to provide it.
Where is your company on the AI adoption curve?
It's still early days — in AI adoption generally, and with AI agents specifically. Key findings from Accounting Seed's 2026 State of AI in Accounting report:
63% — Actively exploring AI
16% — Successfully implemented
Source: Accounting Seed, 2026 State of AI in Accounting — survey of 128 finance professionals Dec'25–Jan'26 (40% from companies with 51–200 employees; 21% from 11–50 employees; 20% from 201–500 employees)
A recent Anthropic study found that fewer than 5% of AI agent tool calls across organizations originate from finance teams — a further indication of how early the adoption of AI agents remains.
Source: Anthropic internal research, 2026
Why AI was a letdown for finance teams — and why that's changing
ChatGPT launched in November 2022 and two months later had over 100 million users. Finance teams tried them. Many came away disappointed — most pilots were abandoned before they ever reached production. The reason is a fundamental mismatch between how LLMs work and what finance professionals expect:
LLMs are probabilistic. They predict the most useful response based on patterns — not running mathematical models or executing formulas. This makes them excellent for researching assumptions, drafting board decks, writing narratives, and generating ideas.
Finance also requires deterministic processes. For financial calculations — a general ledger, a forecast model, a reconciliation — the same inputs must produce the same outputs 100% of the time. These two approaches are not in conflict: the discipline is knowing which to apply and when.
Early tools compounded the problem. Outputs were inconsistent. Numbers were fabricated with confidence. Users didn't yet understand where AI belonged — and where it didn't.
"AI can be remarkably convincing — even when it's wrong. It will tell you it has checked its work. It hasn't, not in the way you mean."
The tools have improved substantially:
Hallucinations have decreased — models are meaningfully more accurate and consistent than they were even 12 months ago
Context windows have expanded dramatically. In 2020, GPT-3 could handle only the equivalent of a few pages at once. By 2026, models like GPT-5 can handle the equivalent of several books in a single session.
Memory is now achievable. Early models had no memory — you had to re-explain everything each session. Now instructions and tasks can be stored as simple text files for repeated use.
We've also moved through three distinct phases, each making the tools more capable — and more consequential:
– Prompt engineering — what we ask
– Context engineering — what information the AI sees
– Agentic engineering — how AI systems plan and act over time
But the lesson remains: use AI where it excels — research, synthesis, drafting, ideation — and use deterministic tools where precision is non-negotiable. Design your workflows to reflect that distinction.
Don't blame the AI model for bad output
Critics are right that AI can produce mediocre and inaccurate outputs — what's now called "AI slop". While AI tools are to blame for some of the errors, users also bear responsibility. Most poor AI results trace back to three causes:
Brief instructions with no context. A two-sentence prompt asking for a financial analysis will produce a generic financial analysis.
An unrealistic accuracy standard. Research on spreadsheet errors by experienced professionals consistently finds error rates of 20–40% in complex models. AI is held to 100% accuracy; humans rarely are
Better prompts, more context, and a verification step produce significantly better results.
3 Key AI Opportunities for CFOs
Understanding how AI applies to finance is now a core CFO competency.
1. Improving internal finance functions to drive growth
Automating routine processes — bank reconciliations, A/R and A/P processing, contract management, variance analysis
Raising output quality — validating forecast assumptions, flagging anomalies, keeping teams current on new accounting standards
Providing better service for revenue operations
2. Providing informed judgment of AI budget requests
Every department is acquiring AI tools. Key questions a CFO should be asking:
What is the total cost of deployment and ongoing usage — including training and ongoing management time?
Do vendors’ AI productivity claims hold up under scrutiny?
3. Understanding the impact of AI on company pricing and costs
• Customers will increasingly recognize that AI lowers supplier costs and ask for price reductions accordingly.
KPMG negotiated a 14% reduction in its audit fee from Grant Thornton on the basis that AI would reduce Grant Thornton's cost of the work.
The reverse is equally true: CFOs should be asking vendors for price reductions at contract renewal — and evaluating how many seat licenses are still needed given AI's ability to reduce headcount requirements.
What implications does AI have on pricing arrangements? If you charge by the user or by the hour, how does AI impact your revenues and how you charge your customers?
Next Issue: Five Principles for Getting AI Right
Issue #2 covers each principle in detail, with examples you can apply immediately.
Start with One Step Today:
Your team is already using AI tools — through official channels and on their own. That's not a problem to shut down. It's information to act on.
Take stock of your AI tools and experiences. Ask individuals and teams what AI tools they're using, how, and for what — and what's working. Understanding where AI is already delivering value is as important as knowing where it isn't.
WORK WITH APERNA
Looking for a fractional CFO — or just AI guidance?
We work with CEOs and finance leaders at small and mid-sized companies. Fractional CFO leadership and practical AI — together.
Reach out to us at [email protected]
Know someone who would find this useful? Forward this newsletter and let them subscribe: newsletter.aperna.com
Brian Hendry, CPA, CFA, MBA · Founder, Aperna Inc.
Fractional CFO advisory for small and mid-sized companies. Toronto, Canada. · aperna.com
AI with Purpose: Drafted and reviewed by humans with the help of Claude AI, Perplexity and NotebookLM.
DISCLAIMER
Professional Disclaimer: The content of this newsletter is provided for general informational and educational purposes only. It does not constitute accounting, tax, financial planning, or investment advice. Aperna Inc. and its principals hold CPA and CFA designations; however, no professional advisory relationship is formed through this newsletter. Consult a qualified professional before making any financial, tax, or investment decisions.
References to third-party companies, products, or services are for illustrative purposes only and do not constitute an endorsement.
Aperna Inc., Toronto, Ontario, Canada.
© 2026 Aperna Inc. All rights reserved.