top of page

What Every Firm Leader Should Understand About Assistive vs. Agentic AI

7 hours ago
5 min read

You can lead your firm through AI without writing a line of code. But you need a clear picture of how AI behaves once it's operating inside your firm. That affects governance, workflows and your risk exposure far more than what tool you choose to implement.


Understanding the difference between assistive AI and agentic AI helps you build that picture. Your team members likely use AI in both ways already, whether they realize it or not.


Assistive AI helps people do the work

Assistive AI helps people work more efficiently by generating suggestions, summarizing information and preparing drafts for human review. It may be embedded in the software your team already uses or available as a separate tool. The defining characteristic is that people decide what actions to take.


It does things like suggesting a revision to an email, summarizing a long client thread, flagging an unusual transaction or proposing a draft response in your tax research platform.


Technology shapes how people work, but people still make the final call.


Because many assistive AI features are built into tools you already use, you may never have formally rolled them out. Instead, a software update introduced a new feature, and your team started using it. You may never have made a deliberate decision to adopt it.


Agentic AI can act on your behalf

Agentic AI goes a step further. Within the authority and boundaries you establish, it can plan steps, use tools and take actions to accomplish a goal without requiring human approval at every step. That doesn't eliminate human oversight or accountability.


For example, an agent might collect client documents, match them to a checklist, follow up on missing items and prepare a return for review. It could reconcile accounts, investigate exceptions and prepare proposed journal entries for approval. It could also draft a client response and gather supporting information, with a team member reviewing the communication before it's sent.


The key distinction is how much authority AI has to act. Assistive AI helps people complete their work. Agentic AI can take action on their behalf within defined boundaries. In either case, your firm remains responsible for the results.


Agentic AI can have a bigger impact on capacity and speed, but you're still responsible for the output.


What changes when AI does the work?

In our Boomer Technology Circle meetings, we often discuss which AI tool firms should implement. But a better question is how AI behaves inside your firm's systems.


Consider the following four areas:

  1. Governance. With assistive AI, you need policies on acceptable use, data handling and review. Agentic AI calls for more. You need to address who approves what an agent can access, what it can change and when a human must sign off. If you can't answer those questions right now, you need to address governance.


  2. Process redesign. Assistive AI speeds up individual steps in your existing workflow. Agentic AI can take over entire segments of that workflow. When an agent does the work, the old handoffs, review points and approval chains might not make sense anymore. You need to redesign the process around what the agent does and what a person must verify.


  3. Role clarity. Your team members need to know what they own as AI handles more production work. A staff accountant who used to prepare reconciliations may now review an agent's output. A manager who used to review everything may now focus on exceptions. Roles shift, and people do their best work when they know what they're accountable for.


  4. Risk exposure. Assistive AI can introduce errors when people rely on inaccurate suggestions without proper review. Agentic AI adds another dimension of risk because it can take action across multiple systems or client engagements. Without appropriate controls, an error can spread before someone catches it. Firms still have professional liability, confidentiality duties and quality management obligations. The greater the autonomy, the more important it becomes to establish clear boundaries, monitoring and accountability.


Systems awareness is more important than tool knowledge

Tools change every few months. The product you evaluate today will look different by this time next year, and new competitors will make big promises to win your business. If you build your leadership approach around knowing the tools, you'll always be chasing the latest release.


Leaders need systems awareness, meaning you understand how work moves through the firm and where people make decisions and need to stay involved. With that understanding, you can evaluate any AI tool, present or future, by asking the same handful of questions.


Start with these:

  1. Does this AI suggest or act? This tells you whether you're managing an assistant or giving AI authority to take action.


  2. What can it access and change? Map the data it reads and the systems it writes to.


  3. Where does a human review the output? Name the person and the point in the process.


  4. What happens when it makes a mistake? Decide in advance how you'll detect, contain and correct errors.


  5. Who is responsible for the outcome? Every AI-assisted process needs a named team member accountable for the result.


If you can't answer these questions about a tool, hold off on rolling it out until you can.


Where to begin

Start with an inventory of the AI features already running in your firm's software, including those your team switched on without asking. Identify where AI assists people and where it can take action on their behalf. Then pick one high-volume process and map it end to end, noting where AI acts, where people review and who owns it.


That exercise shows you where your policies need work and where your processes need redesign. It also gives you a repeatable method, so each new tool fits into a framework you already trust.


AI will continue moving into your firm, often via tools you already use. You need to understand how the technology behaves and build your processes to match. The sooner you start, the more control you have over the results.


Could your firm benefit from getting firm management and IT leaders in alignment?


The Boomer Technology Circles are a peer group of firm and technology leaders in the accounting profession who benefit from aligning IT and firm strategy and building valuable long-term relationships with solution providers and peers. Apply now to start building confidence in your firm's technology decisions.


Jim Boomer is the CEO of Boomer Consulting, Inc., where he helps CPA firm leaders align strategy, technology, and talent to build high-performing, future-ready firms. A respected speaker and facilitator, Jim leads the Boomer Circle communities and speaks regularly at national conferences, including AICPA ENGAGE. Known for his calm presence and strategic mindset, he’s consistently recognized as one of Accounting Today’s Top 100 Most Influential People and IPA’s Most Recommended Consultants. 


Comments


bottom of page