June 8, 2026

AI and humans working together: how the audit team is changing

Amanda Waldmann

Increasing trust with AI for audit and advisory firms.

Key Features:

Work that once scaled mainly through headcount is increasingly being redesigned through AI-assisted workflows.
Partners, managers, seniors, and associates all shift toward judgment, orchestration, and advisory; none of the roles disappear.
Firms that redesign teams intentionally will capture the real economic and retention upside.

Audit teams are under pressure to do more with limited staff, while too much senior time still goes to coordination, review cycles, and manual execution. A growing number of firms are already deploying AI for risk assessment, journal entry testing, and workpaper generation. Every role on the engagement team shifts with it. This article covers the specific audit tasks already being handed off to AI today, how each role on the engagement team is changing, and what firm leaders need to do to get the team design right.

From "Will AI replace auditors?" to "How will we work together?"

Firms are being asked to do more work than the talent pipeline can supply, and even with undergraduate accounting enrollment rising again, hiring alone is not going to close the gap.

AI was never going to replace auditors. But it can change what they spend their time on. Practitioners want to work with AI because the alternative, doing more with fewer people on the same operating model, is not realistic. The interesting question now is how the work gets divided.

The repetitive layer of an engagement, reconciling versions, chasing evidence, formatting workpapers, pulling the same data into the same templates, is where AI does its best work. What stays with practitioners is what the repetitive layer was always crowding out: risk, exceptions, client conversations, and professional skepticism. The team gets redesigned around that split.

Audit tasks already handed off to AI

Three categories of work are already moving to AI on engagements today: risk assessment, journal entry testing, and workpaper standardization.

Risk assessment. AI is analyzing data populations to refine risk identification and audit planning at the front end of the engagement, alongside the sampling approaches required by standards.
Journal entry testing. AI executes the work across large data sets at a scale and speed manual procedures can't match, surfacing the unusual transactions that warrant practitioner investigation.
Workpaper standardization. Firms are standardizing workpaper templates so AI systems have the structure to extract and transform data at scale across varied engagements.

This is the work Fieldguide's Field Agents are purpose-built to execute: planning, evidence review, control testing, and workpaper preparation, with practitioners reviewing the output and applying judgment.

The new audit team: assistive AI, Agent Workforce, and practitioners

Not all AI on an audit engagement is the same, and the distinction matters for how the team is designed. Most AI in the market today is assistive: chat tools, copilots, and point automations that a practitioner triggers and reviews step by step. Useful for individual tasks, but the human still drives every action, so the operating model doesn't really change.

Fieldguide's Agent Workforce is the second category, and the one that actually changes the model. Field Agents execute the engagement work, including planning, evidence validation, control testing, and workpaper preparation. Practitioners direct the workforce through Field Orchestrator, which coordinates the Field Agents, and review every output before it's finalized.

In practice, Fieldguide's Agent Workforce handles the structured execution layer of the engagement, so practitioner time moves to exceptions, client conversations, and the judgment calls only a CPA can make. The economic shift, and the client experience improvement, both come from getting that division right, not from adding another tool on top of the same workflow.

How each role changes

Partner time, manager time, and associate time are not interchangeable, and the role-level shifts are what determine whether the redesign actually lands.

Partner

Partners spend less time on year-end review and sign-off, and more on earlier, governance-oriented involvement. AI governance, including model risk assessments, data lineage, and human-in-the-loop thresholds, is increasingly a firm-wide function with C-suite sponsorship, and partners are central to designing it. The day-to-day shift is what each partner can spend their time on: more client advisory, more governance design, more of the judgment calls that earn the fee. Less review note triage.

Manager

Managers have historically been the coordination and QC layer: chasing status, tracking requests, reviewing workpapers. Field Agents take over the routine testing and documentation, and the role shifts to exception handling, client context, and reviewer judgment.

Senior

Seniors have historically run multi-week execution. With Field Agents executing the documentation and testing layer, the role shifts to overseeing AI outputs and configuring how the work runs. The CPA as AI evaluator is a growing competency: assessing whether AI-flagged anomalies are genuine risks or AI errors is exactly the kind of professional skepticism that distinguishes a strong senior.

Associate

The associate role is where the change is most visible. Most of the work the first three years used to be built around (copying data from PDFs into Excel, reformatting workpapers, chasing missing files) is exactly what Field Agents now execute. What replaces it is harder to train for and more worth doing: pattern recognition, source validation, learning to apply judgment to AI output. For firms competing for the next generation of CPAs, that is the more honest pitch about what early-career audit work looks like.

Uniquely human work on the engagement

Signing the audit opinion is the clearest part of the engagement that stays human. It is a professional act tied to a licensed practitioner, not a task that gets delegated to a system. The judgment calls underneath it, including professional skepticism, client trust, and the ethical responsibility that comes with the credential, are what the opinion ultimately rests on, and they are where an experienced auditor's value is most visible to both the client and the file.

Redesigning the team: practical moves for leaders

Four moves separate teams that capture the upside from teams that just deploy the technology.

Task-level mapping of engagement work

Agents do well with defined work, while practitioners do the work that requires judgment, context, and skepticism. A useful way to map an engagement is to take each task cluster and ask three questions: Is it repeatable? Does it require professional skepticism? Does it consume disproportionate staff hours relative to its judgment content? The repeatable, low-judgment, hour-heavy clusters move to Field Agents, and the higher-judgment work stays with practitioners.

Pre-committing saved hours to higher-value work

A common failure mode is that efficiency gains get absorbed invisibly into existing workloads, and only a small share of organizations are effectively turning AI deployment into business outcomes. Pre-committing reclaimed capacity before deployment, whether to deeper risk coverage, expanded advisory, or new service lines, is what turns hours saved into outcomes the firm can actually point to.

Review checkpoints for AI-generated work

Traditional review structures were designed to catch errors in human-performed work. The new model needs review designed for the work auditors are now doing: confirming scope, validating exceptions surfaced by Field Agents, and checking client-specific risk alignment. That is where senior and manager attention adds the most value, and it is what the Preparer Review Agent is built around.

Performance metrics aligned to the new model

Hours billed stops being a good measure of senior and manager performance once Field Agents are doing the hour-heavy work. The scorecard that fits the new model rewards how well people orchestrate Field Agents, what advisory value they deliver from the time reclaimed, and how clearly they demonstrate professional skepticism on the work that came back from the agents. Agent management is its own skill set, and the firms that get there first will be the ones measuring for it.

Hiring, training, and careers

The hiring profile, the training curriculum, and the retention pitch all shift with the operating model, and firms that update the three together pull ahead of firms that update only one.

Hiring. The profile is moving toward AI-forward generalists who can orchestrate agents and apply judgment to their output. Technical accounting still matters; the addition is comfort directing and reviewing AI-executed work.
Training. CPAs need stronger AI skills, and most still lack a working understanding of how agentic systems behave on an engagement. Less procedural rote, more reviewer judgment and agent supervision.
Retention. The career path argument is the strongest one firms have. An engagement model where associates spend their time on pattern recognition, exception investigation, and client interaction is more sustainable than 60-hour weeks of manual, repetitive work, and easier to retain people through.

Accounting enrollment is rising again, and the firms that can tell incoming talent a credible story about what the first three years actually look like will win the recruiting cycle.

Build the team for the next decade of engagements

Fieldguide is the industry's only end-to-end AI-native platform purpose-built for audit and advisory, with the Agent Workforce, methodology depth, and audit-grade rigor firms need to operate the new model. Field Agents execute planning, evidence review, control testing, and workpaper preparation across the engagement lifecycle, while practitioners direct the work, review the output, and retain professional judgment on every finding and conclusion. That is what makes team redesign real instead of theoretical: a production-ready foundation that firms can build their differentiation on top of. Request a demo to see how the new model works in practice.