June 15, 2026

How Agentic AI Reshapes Which Engagements a Firm Takes On

Amanda Waldmann

Increasing trust with AI for audit and advisory firms.

Key insights:

Once agents handle the procedural middle, you can rethink which work the firm takes on, not just how fast it ships.
The talent gap is structural. The recent enrollment bump is real, but it won't reach your bench in time.
Agents work as evidence arrives, so busy season stops being the only mode the firm runs in.
The real win isn't speed. It's being able to reprice or drop low-margin work and put senior time on advisory.

Most firms today run engagements the same way they ran them 10 years ago. A senior associate spends the first hour of the day pulling the prior-year binder, the next two reconciling what the client uploaded against what was actually requested, and the rest of the week renaming PDFs and chasing missing items. That model worked when there were enough people to staff it. It doesn't anymore.

The fix most firms reach for is AI that helps people move faster through the same steps. That isn't the shift that matters. The real change comes when agents do the procedural work themselves and the firm gets to decide what to do with the capacity that frees up. This post looks at what that changes: which engagements are worth taking on, how busy season plays out, and what it means for the profession as the model spreads.

What is an agentic AI workforce in audit?

Agentic AI systems execute tasks on a user's behalf and differ from simpler AI assistants that respond step by step to prompts. They pursue multistep goals across an entire workflow, executing the procedural steps while practitioners set direction and review the results. AI assistants simplify tasks; agentic systems complete them.

That's the ceiling most AI tools in audit run into, because most of them are assistants. You ask a question, get an answer, and decide what to do with it. That's useful, and it speeds up research and document review. But the underlying engagement model is unchanged: a human still initiates each step, executes it, and decides what comes next. A firm can roll out copilots and chat tools across the team and still watch realization stay flat, because the workflow around those tasks hasn't moved.

That's the line Fieldguide draws with its two categories of AI. AI Assist gives practitioners on-demand AI for the tasks they want to drive directly. The Agent Workforce takes on the multi-step procedural work end-to-end, with practitioners reviewing. Both run on the same platform, and the rest of this post follows what the Agent Workforce changes.

How agents and practitioners split the work on an engagement

Inside an engagement, the Agent Workforce runs on a clear chain. Practitioners direct the work, Fieldguide's Field Orchestrator coordinates the workflow, and Field Agents execute within the parameters the practitioner sets. The agents take the sequential, evidence-heavy work that eats the most hours. They can support:

Review of uploaded client evidence for completeness and readiness
Flagging missing items and inconsistencies early
Test procedures by matching evidence to samples
Identifying exceptions
Documenting results for practitioner review

AI Assist sits alongside that for the work a practitioner would rather run directly. AI Actions, for instance, generates outputs across a column of a sheet in one click, with multi-step reasoning across each row, for repeated tasks like drafting and summarizing.

The division of labor is the point. Agents handle the volume; the practitioner reviews, judges, and signs.

How an agent workforce changes audit engagement economics

Most AI conversations in audit and advisory stop at "engagements get faster." That framing misses the bigger lever.

Which engagements you take on

When Field Agents own the procedural middle of an engagement, the math on which work the firm takes changes. The compliance work that today consumes senior hours because there is no one else to do it can be priced and staffed differently. A lower-margin recurring attest engagement that the partner has tolerated for years because the team capacity it absorbs is hard to redeploy becomes a real choice rather than a default. Walking away from work that doesn't fit the firm's positioning, or pricing it at what it actually costs, becomes a portfolio decision.

The shift toward advisory

Routine attest and compliance work stops monopolizing senior capacity, which opens room for a higher proportion of advisory-heavy work where partner judgment carries the value. The partner whose calendar today is half-consumed by review notes on procedural workpapers can spend that time on the conversations that lead to advisory expansions.

Faster expansion into new service lines

Expanding into a new service line today means a hiring plan, a training plan, and a staffing plan that takes two to three years to mature. Agent configurations move at a different speed. A practice leader who needs to surge into a new regulatory framework, a new industry vertical, or a one-off engagement type can move faster than a full hiring cycle allows, provided the firm still has practitioners who hold the framework expertise to direct and sign the work. Service lines can expand and contract closer to the speed of client demand.

This is the conversation most competitors don't structure for partners. "X% faster testing" is a feature claim. Reconfiguring revenue mix is a firm strategy.

How agentic AI changes busy season for audit firms

An agentic workforce can change when work happens, not just who does it.

Today's engagement calendar runs on spikes. The model assumes a team can scale its hours up two- or threefold through busy season, January to April, then ease off. Most of the attrition data suggests they can't keep doing it. The senior and manager ranks absorb the worst of it, regularly logging 60-plus-hour weeks in the busiest stretch, and fatigue at that level is a known drag on audit quality, which turns a staffing problem into a review-risk problem. Worse, the people who burn out tend to be the experienced ones the firm can least afford to lose, and replacing them costs more than keeping them would have. The spike doesn't have to be the only way the work gets done.

The work starts before the team does. A client uploads the trial balance and supporting documents in late January, and Field Agents check completeness, compare against prior-year mappings, flag missing items, and stage the evidence against the testing plan. By the time the senior associate sits down on Monday, the cash and AP sections already have evidence matched and exceptions flagged, so the week opens on review instead of setup. The peak doesn't vanish, but the team isn't trying to carry all of it in the same compressed weeks.

Work also stops stacking up between engagements. When a client changes systems or restructures a process mid-year, Field Agents can refresh the affected testing and documentation then, instead of leaving it to pile up for next year's fieldwork, so the next full engagement starts closer to ready. For the staff in the middle of the firm, that is what bends the hours curve down. The 80-hour weeks that push people out are downstream of procedural crunch, and moving the procedural work off the spike is what shortens them.

What an agent workforce changes across the profession

Most writing on AI in audit assumes a single firm boundary. The bigger shift shows up when an agentic workforce becomes the operating layer the profession itself runs on.

Mergers and cross-border work

Integration friction is rarely the methodology document; it's the gap between what the methodology says and how each team actually executes it. An agentic workforce that runs the firm's methodology the same way across every engagement closes that gap. When a firm acquires a practice or coordinates across borders, the shared workflow is already consistent, and people handle the local regulation, client nuance, and judgment that doesn't standardize.

When senior people move

A partner's value has always included tacit knowledge: how they scope, where they push back, what they document and what they don't. In an agentic model, part of that lives in how they configure and supervise Field Agents, so it doesn't all walk out the door when they leave. That portability changes how firms think about retention, succession, and lateral hiring.

Consistency across jurisdictions

Audit work fragments to match each jurisdiction's quirks, and the variance shows up in the file. Standardized agent workflows hold the procedural baseline steadier across different regulatory environments, so the variance that remains sits in the human judgment layer where it belongs.

The accounting talent shortage driving AI adoption

These shifts used to be optional. The talent shortage is what makes them necessary now, and the math behind it hasn't gotten better.

Total accounting degrees dropped 6.6% in the 2023–24 academic year, part of a multi-year decline. Fewer graduates means fewer new CPAs entering the pipeline, and the constraint shows up first in the staff and senior ranks that carry the procedural load on every engagement.

Demand runs the other direction. The BLS projects 5% employment growth for accountants and auditors from 2024 to 2034, with about 124,200 openings projected each year, many of them to replace people who retire or leave the field. A shrinking pipeline feeding a steady stream of openings is the gap firms are staffing against.

Signs of recovery, but not enough

There are positive signals: fall 2025 enrollment reached 313,397 students, up from 293,759 the prior year. But enrollment gains take years to convert into licensed, experienced practitioners. A student entering an accounting program today still faces:

The 150-credit-hour requirement in most states
CPA exam passage
One to two years of supervised experience before qualifying for licensure

That pipeline runs four to six years before a new graduate becomes a practitioner who can carry real responsibility. Firms need that capacity now, not at the end of the decade. The only lever that moves on the right timeline is changing how the work gets done, so the people already on staff can take on more without burning out.

Build your agent workforce on Fieldguide

Firms adopting an agentic workforce today are changing how engagement work gets done. In 2026, Fieldguide will have a full set of Field Agents across the entire engagement workflow, as well as supporting analytics and orchestration, live in production. Fieldguide AI gives firms one place to coordinate engagement work, reshape portfolio mix, and ease the busy-season spike, with practitioners directing the work and signing off at every stage. Request a demo to see how it maps to the way your firm runs engagements.