Advisory and audit solutions provider
Fieldguide CEO Jin Chang, in an interview, said the term "AI" has become very broad, which has caused some confusion at firms that generally understand they need to adopt AI but aren't entirely sure what that means specifically. He noted that some of this confusion comes from ambiguity over the definition of terms like "generative," "agentic" and even "AI" itself.
"What we hear in the market is just so much noise around AI and, nowadays, agentic AI. Every vendor is saying they have agentic AI and, if you ask me, most [claims] are not true. … And we see the same thing happening with [optical character recognition] versus generative AI. Yes, OCR is valuable, but it's not generative AI. If anything, generative AI is better at OCR than OCR!," he said.
The AI Maturity Framework is meant to address this problem by focusing less on specific terms with ambiguous meanings and more on specific capacities for firms. The basic framework is composed of six levels:
- Level 0 – No automation: Firms remain reliant on manual processes, with practitioners burdened by spreadsheets, emails and disconnected systems. Capacity is capped, quality is inconsistent, and growth depends solely on hiring.
- Level 1 – Basic automation: Productivity tools begin to save time, but workflows are still fragmented. Firms see marginal efficiency gains, yet practitioners remain operators rather than advisors.
- Level 2 – Assisted automation: Purpose-built AI embedded in workflows accelerates key steps of the engagement. Practitioners reclaim time for review and insight, beginning the shift from executors to orchestrators. Engagements become more efficient and consistent.
- Level 3 – Directed automation: AI agents manage segments of engagements with human checkpoints. Firms start to scale capacity, while practitioners move into orchestration and exception handling, focusing their expertise where it matters most.
- Level 4 – Guided automation: AI carries the majority of execution. Humans supervise, interpret results and guide client strategy. The firm gains both speed and consistency, while practitioners expand their role as trusted advisors.
- Level 5 – Strategic automation: AI manages entire lifecycles with adaptive intelligence. Professionals focus on foresight, innovation and stewardship of trust. Firms achieve scalable growth, practitioners shape the future of the profession, and clients receive deeper, more strategic value.
So, rather than firms thinking in terms of specific solutions that may be "agentic" or "generative" or something else, they'll focus on process automation, workflow dispersal and the role of human professionals. Each stage is designed to help firms gradually adopt AI in a controlled, strategic manner, with the goal of making meaningful progress without sacrificing quality, trust or oversight. Chang said this framework came from conversations with firm leaders, all of whom were working hard to separate hype from reality when it came to AI.
"Every CPA firm leader we spoke to was saying 'I know I need to invest in AI, but everyone is saying they're AI, help me cut through the noise.' … What we started to see is, depending on where a firm is, there is a different set of things that might need to be accomplished for their AI transformation journey," said Chang, pointing out that some were very early in their AI transformation while others were very advanced. "That's why we started to map out into different levels of autonomy, level zero through five."
One effect of this might be that a firm finds out it's not as AI-enabled as it might think. Chang said that when someone uses an off-the-shelf AI solution or a public model like ChatGPT, they're not really getting the best value from these tools, as they're not specifically built for the profession. And even if they go beyond that and do invest in AI software, it's not much better if all they do is get point solutions that are useful only in very specific circumstances. In these cases, a firm might have trouble grasping the real value that more complex, specialized solutions can bring, said Chang.
"We just felt like the industry needed a bit more clarity around how to approach their AI transformations. … In a professional setting, we need more rigor, a higher bar for quality, more attention to security and privacy practices, and a really actionable framework to help guide CPA firms through their journeys," said Chang.
Fieldguide has released this framework to the world, hoping it will serve as a guide for the entire industry. The goal is less to serve Fieldguide itself and more to drive industry alignment so people can better understand "what true agentic AI looks like." If other vendors use it as a guide to better clarify their own terms and help the industry converge on "their expectations toward what a true agentic platform is," that's a good thing.
"First and foremost, we want the AI maturity framework to exist on its own in the industry as a helpful guide for your firm out there," said Chang.
To download the framework, visit
