Related posts
See all
Key Insights:
- An AI strategy sets a sequenced plan across practice areas, governance, and outcomes.
- Only 8% of senior finance leaders say their organization is well prepared for AI, creating a competitive window for firms that move with intent.
- Governance and data standardization have to move alongside the first engagements from day one.
Most firms have an AI story by now. A few pilots running, a tool or two rolled out, a slide deck that talks about transformation. What most firms don't have is a strategy: a clear view of which practice moves first, what outcomes justify the move, and what governance has to hold before anything scales beyond the pilot team.
That gap is where the real work sits. The firms moving first aren't the ones with the longest tool list. They're the ones who decided, on purpose, where each practice area should be in the next twelve months and built the foundations to get there.
This article covers how to read where your practices actually stand, anchor the strategy to outcomes that justify the investment, and build governance and data foundations in parallel with the workflow plays.
Strategy beyond "buy tools"
Picking tools is the easy part. The harder question is whether the firm has a shared direction for how those tools fit into the engagement lifecycle, the methodology, and the governance model. Without it, each practice ends up optimizing for its own pilot, on its own timeline, against its own definition of success. The result is activity without an operating model.
There are also broader risks that show up. Across industries, AI initiatives consistently stall between pilot and production when data quality, risk controls, costs, and business value are unresolved. An AI strategy helps audit and advisory firms avoid those pitfalls by giving each practice a target stage, a governance structure that holds across the firm, and a sequence that determines what moves first and how fast.
Why AI strategy is really an operating model decision
Engagement work is moving to a new operating model where practitioners and AI agents work together on every engagement. Agents handle more of the execution. Humans lead the review and the judgment. Every AI decision a firm makes, from a single pilot to a multi-year roadmap, is really a call about how fast each practice gets there.
That shift changes where the time goes. Senior staff spend less of it pushing work through and more of it on review, judgment, and client conversations. Junior staff develop differently, because they aren't grinding through the same volume of manual tickmarks. The economics shift too. Hours-per-engagement drops, and the conversation about realization, pricing, and capacity reopens. A strategy is what makes those shifts happen on purpose.
1. Assess where each practice actually stands
A firm-wide AI strategy starts with an honest read of where each practice sits today. Most leaders already sense that their risk advisory, financial audit, and compliance teams are at different points on the curve. Naming those stages explicitly is what makes the strategy actionable.
Map each practice to an AI maturity stage
Fieldguide built the AI Maturity Framework to define six levels of autonomy a practice can sit in, from no AI in engagements at all to an integrated operating model where agents and practitioners run engagements together. It gives leadership a shared vocabulary for placing each practice on the curve. The goal isn't to grade anyone; it's to make explicit where each practice is today and where it needs to go next.
Account for the cost of stage gaps
When one practice operates at a fundamentally different level than another, the firm cannot share methodology, governance, or staffing models across groups. Every new engagement becomes a one-off instead of a repeatable pattern, and the firm loses the compounding benefit a unified strategy delivers.
2. Anchor the strategy to specific outcomes
The outcome the firm is trying to drive should determine every other decision in the strategy: which workflows move first, how aggressive the rollout looks, and how success gets measured. The outcomes that justify an AI investment usually sit in five buckets:
- Capacity expansion
- Margin improvement
- Audit quality
- Talent retention
- Client experience
Pursuing all of them at once diffuses focus and makes it impossible to measure whether the strategy is working. Picking one or two as the primary anchor doesn't mean ignoring the rest; it means knowing what to optimize for when trade-offs come up.
A capacity-driven strategy looks different from a quality-driven one. Go for capacity, and the workflows that move first are the ones eating the most hours per engagement: evidence intake, request management, controls testing. Go for quality, and exception flagging and documentation come first, with a longer ramp in exchange for tighter risk controls. The sequence falls out of the outcome.
Build the business case for recovered capacity
A lot of firms still have no plan for the capacity AI frees up. If you can't articulate what you'll do with the recovered hours, whether that's taking on more engagements, building out advisory, or just letting your people breathe, the full potential of the AI investment won't be realized.
3. Choose the workflows and the ambition level
With the outcome set, the next decision is where to start and what to run the work on. The firms making the fastest progress aren't piloting AI on the edges of the engagement; they're putting AI on the workflows that matter most.
Put AI on the workflows that matter most
The biggest returns come from the workflows that consume the most hours: client request and evidence intake, controls testing, and reporting. Fieldguide's Field Agents execute end-to-end across those workflows, with practitioners reviewing the output. Capacity and quality both move at once, on the workflows where they were under the most pressure in the first place.
Bolt-on AI vs. agents embedded in the engagement
Most AI in audit today is bolt-on: generic chat tools, copilots, and point automations that sit alongside the engagement instead of inside it. Practitioners still reconcile outputs, chase data, and act as the system integrator between disconnected tools. Fieldguide's AI is embedded in the engagement context from the start, and the Agent Workforce is where the operating model actually changes: Field Agents execute structured procedures end-to-end, and practitioners review and apply judgment. Ambition should reflect the practice's risk profile, but the destination is the same: every practice running on the model.
Build vs. buy is the wrong question
Firms with engineering talent reach instinctively for the build option, and the instinct isn't wrong. You want something tailored, differentiated, and aligned to your methodology. The problem isn't the initial build; it's the permanent operational commitment behind it. Security patches, framework upgrades, model updates, integration maintenance, and AI orchestration sit on the same team that's supposed to be shipping new features. AI capabilities also evolve weekly, and an internal team will always be choosing between keeping up and moving forward.
The firms moving fastest aren't picking a side. They partner for the commodity infrastructure (model orchestration, agent architecture, integrations, the things every firm needs and no firm wins on) and reserve their engineering capacity for what actually differentiates the firm on top of that foundation: custom agents, firm-specific methodology, and the workflows that make their service unique.
4. Build governance and data foundations in parallel
Governance, data readiness, and platform decisions can't wait until the first engagements are running. They have to move alongside the workflow plays from day one. The pattern across industries is consistent: of the companies planning to deploy agentic AI in the next two years, only 21% report a mature governance model for AI agents, and at least 50% of generative AI projects are abandoned after proof of concept due to poor data quality, inadequate risk controls, escalating costs, or unclear business value. Audit and advisory firms have an advantage here: methodology depth and regulated workflows mean the governance discipline is already in the firm's DNA. The opportunity is to extend it to AI.
Build governance that holds up to inspection
Governance matters earlier for audit and advisory firms than for most industries because inspection is part of the job. The PCAOB and SEC are both actively developing guidance on AI use in audits, but comprehensive AI-specific standards are still taking shape. Firms that document AI usage, accountability, and review steps proactively are in a much stronger position than firms explaining ad hoc usage during an inspection.
Standardize methodology before automating it
AI outputs are only as consistent as the methodology underneath them. Knowledge bases, testing approaches, and documentation standards have to be captured somewhere agents can draw on them; on the right platform, that capture happens as the work runs. Skip the standardization entirely and AI outputs vary engagement to engagement because the underlying methodology was never made explicit.
5. Measure, codify, and scale on real engagements
With outcomes anchored, workflows chosen, and governance in motion, the strategy starts producing real work. The first engagements on the right platform aren't a science experiment. They're real client work, with the team learning what to keep, what to tune, and what to take across the rest of the practice. The mechanics matter: the right metrics, a clear path from what works to a repeatable playbook, and a plan for scale before the first engagement goes live.
Measure against the chosen outcome
Metrics should map directly to the outcome the strategy is anchored to. Capacity-driven strategies measure hours per engagement. Quality-driven strategies track exception rates and cycle time. Margin-driven strategies look at realization and write-off. Without that anchor, the early engagements end with a deck of activity metrics and no clear answer on whether to scale, adjust the configuration, or take the next practice live.
Turn what works into a playbook
What works has to harden into a playbook. Methodology settings, agent parameters, checkpoints, documentation standards: each new engagement starts where the last one finished instead of starting from scratch. Repeatability is where the gains compound.
Plan for the hard part: scale
Scale is where most strategies stall. The translation from "this worked on a few engagements" to "this is how we run the practice" requires leadership commitment, governance infrastructure, and a platform that supports the full lifecycle.
6. A realistic adoption cadence
Firms can be running real engagements on Fieldguide within weeks, not months, with custom capabilities layered on over time. The arc below assumes that pace:
- Weeks 1–4: Anchor and assess. Outcomes get picked, each practice area maps to its current stage on the maturity curve, and the first one or two workflows come into focus.
- Weeks 4–8: Launch the first plays. Agents get configured to the firm's methodology, initial engagements run on the platform, and governance and data standardization start in parallel.
- Weeks 8–16: Codify and expand. Playbooks form from what worked, the governance structure finalizes, and additional engagement teams within the same practice area come on.
- Months 4–6: Scale and set the next ambition. The proven model rolls across the practice. Stages get reassessed. Targets are set for the next practice area and the next level of ambition.
The shape is consistent: anchor, launch, codify, scale. The firms that climb the curve are the ones running the sequence on a platform built to support it.
Run your AI strategy on a platform built for the work
A strategy only delivers if it runs on infrastructure built for audit and advisory work. 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 move from isolated pilots to a repeatable operating model across every practice area. Fieldguide's Field Agents execute across the engagement workflow, with analytics and orchestration alongside. All AI outputs still require practitioner review and professional judgment. To see how the strategy comes to life on the platform, request a demo.
Amanda Waldmann
Increasing trust with AI for audit and advisory firms.
