Figuring out your tech stack as a small or midsize firm is essential to staying competitive in the marketplace. But what does it actually look like to adopt AI tools? How can you choose the right vendors? And most importantly, how can you keep up with market trends without sacrificing client relationships or security?
In our latest report, we partnered with CPA Crossings to explore AI adoption specifically as it relates to the needs of small and midsize CPA firms. Here are the key takeaways from “Start from Trust: Adopting AI and New Technology in Small and Midsize CPA Firms.”
Use the Trust Framework for Better Technology Decisions
Tax and compliance work is built upon a foundation of trust. Dealing with confidential financial information requires it. When evaluating new technologies, it’s a no-brainer to incorporate this same framework into your firm’s decision-making process. Firms should use their instincts when it comes to AI adoption—the same instincts they use to protect client information. Technology decisions take place in the context of client relationships, professional judgment and regulatory duty.
Validate Skepticism
Not everyone is eager to adopt AI into their workflows, especially when it comes to smaller firms. In a 2024 CPAFMA IT survey, 63% of responding firms selected “Neutral/Wait and See” in response to the question of how aggressively they planned to adopt AI. Meanwhile, just 28% of participating firms described themselves as excited or hopeful. Only 22% had already implemented an AI policy.
Further, data from an AICPA/CIMA survey published in the Journal of Accountancy found that 71% of finance executives were at least moderately concerned about privacy, ethics or accuracy risks. The survey also found that only 6% had implemented generative AI in one or more business applications in the first quarter of 2024. By the end of that year, implementation remained the same at 6%, despite growing interest.
When firms have an AI policy in place, data stays safe and learnings are documented across the organization.
This caution around experimentation is healthy, and the firms who are holding off aren’t doing so because they don’t get it. They’re doing so because they want to ask the right questions and understand the safety, risks and benefits of incorporating AI.
Consider the Spectrum of Adoption
Small and midsize firms operate on a spectrum when it comes to opinions on AI usage. Some firms consider it when hiring new talent and evaluating employee competencies. But many smaller firms don’t consider it at all. In our report, we found 22% ranked it the single most important nonaccounting competency and an equal percentage ranked it the least important.
Curious to see where your firm falls? Consider the spectrum below:
Failure Mode: Delay
- Paralysis: fear of doing AI wrong, which leads to not asking questions.
- Accidental delay: unintentionally avoiding a decision.
Defensible Ground
- Informed caution: Questions are asked without assumptions.
- Responsible piloting: Pilots are put in place, results are measured with human oversight.
Failure Mode: Recklessness
- Reckless adoption: Technologies are adopted quickly and without scrutiny.
Using AI in an Unstructured or Governed Way
Whether an organization has a policy in place or not, AI is likely already being used within a firm. A common problem for small and midsized firms without an AI policy or tools in place is that many employees will use their own personal tools. This can create security risks. When firms have an AI policy in place, data stays safe and learnings are documented across the organization.
Evaluate Vendor Claims
When small and midsized firms do choose to adopt AI, doing intensive research on different vendors is essential. Blind adoption can lead to wasted capital, privacy risks and no results. It’s important to keep in mind that just because there’s a lot of buzz around a vendor, doesn’t mean they can live up to the hype and be what your firm needs.
Firms should ask several key questions when evaluating a vendor:
- Can the tool inherit your firm’s permission structure so the AI cannot “see” more than the human user is allowed to see?
- Can you tell when the system is uncertain, and does it trigger human review instead of bluffing through a result?
- Are there audit trails for AI-generated actions or suggested conclusions?
- Can you explain to a client, regulator or internal reviewer what the tool did and where judgment still sat with the practitioner?
- If a model changes, can the vendor describe testing, rollback and incident response procedures?
These questions incorporate the trust framework and connect AI vendor evaluations with the way CPA firms already handle risk.
You can also prevent a bad vendor match by looking out for red flags. Can a potential vendor answer CPA-specific questions with more than generic SaaS language? Do they pressure a contract prior to a real pilot?
Small and midsize firms can benefit greatly from evaluating their current viewpoints around AI and approaching it with the same principle they approach their work: trust.
Read the full report and gain more insights: “Start from Trust: Adopting AI and New Technology in Small and Midsize CPA Firms.”
