What the Coming Agentic Workforce Means for Professional Services Firms in USA

Direct answer: A significant agentic workforce — AI systems that plan, execute, and complete multi-step work with minimal human handoff — is arriving inside professional services firms faster than most of those firms are organizationally set up to absorb. For US law firms, accounting practices, consulting shops, and similar knowledge-work businesses, the near-term risk isn't that agents replace people outright; it's that firms without clear governance, workflow redesign, and client-facing readiness fall behind competitors who integrate agents deliberately. Firms that treat this as an operating-model question now, rather than a tooling question to revisit later, are the ones positioned to come out ahead.

Industry commentary from longtime marketing and media strategist Rishad Tobaccowala, circulating in August 2026, has put a name to a concern many operators have been dancing around without saying it plainly: a genuinely significant agentic workforce is coming, and it is coming faster than the org charts, approval chains, and client-service models of most companies can absorb. That commentary doesn't attach a specific adoption percentage, a named client roster, or a dollar figure to the claim, and neither will this piece — a precise, verifiable statistic for how many US professional services firms have deployed agentic systems today isn't publicly available at this level of specificity, so what follows reasons from the shape of the trend rather than a number nobody can responsibly cite. What is clear, and doesn't need a footnote to be credible, is the underlying pattern: agentic AI capability is now good enough to touch real client work — drafting, research synthesis, intake, scheduling, first-pass document review — well before most firms have decided who owns that capability, how it gets audited, or what happens when it makes a mistake in front of a client. That gap between what the technology can do and how ready the organization is to govern it is the actual story, and it lands hardest on professional services firms because their entire business model runs on billable, judgment-heavy human labor. The rest of this piece works through what the shift actually looks like inside a US professional services firm, why the timing problem is sharper here than in most sectors, and what changes in your website, client portal, and internal workflows before the gap becomes a competitive liability instead of a manageable transition.

What "Agentic Workforce" Actually Means

The phrase gets used loosely, so it's worth being precise about what separates an "agentic workforce" from the chatbots and copilots most firms already have running somewhere in their stack. A copilot suggests; a chatbot answers a single question inside a single turn. An agent does something different: it takes a goal, breaks it into steps, calls tools or systems to execute those steps, checks its own output against the goal, and only surfaces to a human when it hits a decision point it isn't authorized to make on its own. The distinction matters because the risk profile is completely different. A chatbot that gives a wrong answer produces one bad response a person can catch. An agent that mishandles a multi-step process — say, pulling the wrong precedent into a client memo, misrouting a compliance document, or auto-scheduling a meeting with the wrong counterparty — can propagate an error across several downstream actions before anyone notices.

This is also why the word "workforce" in the commentary is doing real work rather than being a figure of speech. A single agent handling one task is a productivity tool. A set of agents that hand work to each other — one drafting a first-pass summary, another checking it against firm templates, a third routing the result to the right reviewer — starts to resemble a small team operating in parallel with human staff. That's a different planning problem than buying software: it means a firm needs to decide where in its actual org chart this "team" reports, who signs off on its output, and how its performance gets reviewed, the same categories of decision a firm would make for a new human hire, compressed into a much shorter adoption timeline.

From Copilots to Coworkers

The shift Tobaccowala's commentary describes isn't "AI gets a bit better." It's a category change: from tools that sit inside a person's workflow to systems that carry out a slice of the workflow independently, the way a junior associate, paralegal, or research analyst would, minus the years of onboarding and minus a badge that opens the building. That's a meaningful capability jump for professional services specifically, because so much of the work in these firms — legal research, tax preparation, financial modeling, market analysis, contract review — is precisely the kind of structured, document-heavy, rules-based knowledge work that agentic systems are best at handling today. It's not that agents are uniformly better than experienced professionals; it's that they're now capable enough on a meaningful subset of tasks that ignoring the category is no longer a neutral choice.

Why This Hits Professional Services Firms in the USA Specifically

Every industry is dealing with some version of agentic AI arriving on a faster timeline than internal readiness. Professional services firms feel it more acutely for a structural reason: their revenue model is built directly on hours of human judgment, and their client relationships are built on trust that a licensed, accountable professional stood behind the work. When a meaningful share of the underlying tasks can be executed by an agent, both pillars get pressure-tested at once — the billing model and the trust model.

In the US market specifically, this collides with three forces that are already active. First, clients of law firms, accounting practices, and consultancies are themselves adopting AI internally and are increasingly asking their outside firms pointed questions about turnaround time and pricing that assume some automation is already happening. A client who knows their own team can draft a first-pass contract summary with an agent will not indefinitely accept paying full hourly rates for a human to do the identical first pass. Second, US regulatory and liability exposure — malpractice standards, SEC and IRS scrutiny, state bar rules on AI-assisted work product — means the risk of moving too fast without governance is real and specific, not hypothetical. Third, competitive dynamics in US professional services are unusually visible: firms compete regionally and nationally for the same corporate clients, and a competitor that quietly cuts turnaround time in half on routine matters becomes a pricing and speed threat before anyone else notices why.

The Talent and Margin Squeeze

There's also a talent dimension that's easy to miss. The entry-level work historically given to first- and second-year associates, junior accountants, and research analysts — the repetitive, high-volume, lower-judgment tasks — is exactly the work agentic systems are most capable of absorbing right now. That creates a genuine strategic question for US professional services firms: if agents handle the bulk of the traditional training-ground work, how does the firm develop its next generation of senior judgment, and how does it restructure the associate-to-partner pipeline that the whole business model depends on? This isn't a problem a single software purchase solves. It's an organizational design question that firms need to be actively working through now, not reacting to after headcount decisions have already been made under pressure.

What Changes in Practice for Your Website, Client Portal, and Workflows

Set the philosophical questions aside for a moment and look at what actually has to change on the ground — starting with the systems clients touch directly. A firm's website and client portal are usually the first place agentic capability shows up in a form clients can see, because intake, scheduling, document collection, and status updates are structured, repeatable processes that map cleanly onto what an agent does well. Done properly, this is where firms find genuine near-term value: an agent that qualifies inbound leads against your practice areas, schedules a consult with the right partner, and pre-populates an intake form from a client's own documents removes hours of coordination work without touching the judgment-heavy parts of a matter. This is the practical starting point most firms should evaluate first, and it's the kind of build that falls under a firm's AI Agents & Automation work rather than a generic chatbot widget — the difference being that an agent actually completes the intake and scheduling sequence end to end instead of just answering questions about it.

Client-Facing Risk Surfaces You Didn't Have Before

Every new automated touchpoint on a client-facing system is also a new surface for something to go wrong, and professional services firms are not used to thinking about this the way ecommerce operators have had to. If your firm accepts retainer payments, invoice settlements, or deposit collection online — and most do now — an agent-driven intake or billing flow introduces the same category of exposure that online retailers have spent years hardening against: synthetic identities at intake, disputed retainer charges, and account-takeover attempts against a client portal that now moves faster and asks fewer manual verification questions than it used to. The patterns are close enough to what's covered in Ecommerce Fraud Prevention: Protecting Your Store From Chargebacks that it's worth treating client-portal payment flows with the same seriousness once an agent is initiating or approving parts of that sequence rather than a staff member who would have caught an obvious anomaly by instinct.

Marketing and Visibility Can't Pause While You Sort This Out

The organizational work of getting ready for an agentic workforce doesn't happen in a vacuum — client acquisition keeps running while it does, and professional services firms shouldn't treat marketing as a lower priority just because the bigger strategic question is elsewhere. Two adjacent shifts are worth folding into the same planning cycle. Firms explaining a new service line, a practice area, or how they're using AI responsibly are finding that short explainer video content converts better than static pages for this kind of trust-building message, and the production process for that has changed enough that it's worth reading through AI Video Ads: How to Create Them (2026 Guide) before commissioning anything the old way. Separately, US professional services firms with more than one office — a common structure for regional law firms, accounting practices, and consultancies — are still leaving visibility on the table if each location doesn't have its own properly built local page; the approach in SEO for Multi-Location Businesses: Local Pages Done Right applies directly to a firm with, say, a Chicago and an Austin office competing for local search visibility in each market independently.

Where This Work Typically Falls in Terms of Investment

Firms asking "what does it cost to actually do this" are usually better served by thinking in terms of scope tiers than a single number, because an intake agent for one practice group is a very different build than a firm-wide rollout across matter management, billing, and client communications. As a general reference point for what this kind of work typically falls under:

Tier	Typical scope	Starting investment
Essential	A single agentic workflow — client intake, scheduling, or first-pass document sorting for one practice area	$1,000
Growth	Multiple connected workflows across intake, portal, and internal document handling, with basic governance and audit logging	$2,000
Enterprise	Firm-wide agentic rollout across practice groups, integrated with existing case/practice management systems, full audit trail and compliance controls	$4,000+

These are starting points, not fixed quotes — the right scope depends on how many systems the agent needs to touch and how much governance the firm's compliance obligations require. The point of the table is directional: most firms should be planning around a phased Essential-to-Growth path rather than either doing nothing or attempting an Enterprise-scale rollout before they've proven the model on one workflow.

How to Get Organizationally Ready Before the Agents Arrive

Tobaccowala's framing is useful precisely because it separates two things that get conflated constantly: technical readiness and organizational readiness. A firm can license every agentic tool on the market and still be unready if nobody has decided who is accountable when an agent's output goes into a client deliverable, how errors get caught before they reach a client, and what training changes for staff whose junior-level tasks are now partially automated.

Practical readiness starts with governance that exists before deployment, not after an incident. That means a documented policy on which tasks an agent is authorized to complete unsupervised versus which require human sign-off, an audit trail that records what an agent did and why for every client-facing action, and a named owner — not a committee — responsible for the agent's performance the way a practice group leader is responsible for associate work product. It also means rethinking training: if agents absorb the repetitive research and drafting tasks that used to teach junior staff the fundamentals, firms need a deliberate plan for how those staff still develop judgment, because that judgment is exactly what clients are paying a premium for once the repetitive work is automated. Firms that skip this and simply lay off the junior tier without replacing the training function are trading a short-term cost saving for a multi-year gap in their senior bench.

None of this requires a firm to have every answer before it starts. It requires the firm to have decided, in writing, who answers those questions as they come up, and to have built a feedback loop where staff who spot an agent error have a clear, fast way to flag it and see the correction actually made. Firms that leave this informal tend to discover the gap only after a client-facing mistake has already happened, at which point the fix is reactive rather than designed.

A Practical 30/60/90 Rollout Path

Given the timing pressure the source commentary describes, a firm doesn't have the luxury of a year-long study before acting, but it also shouldn't deploy agents into client-facing workflows without sequencing the rollout deliberately.

In the first 30 days, the priority is mapping, not building: identify which client-facing and internal workflows are structured and repeatable enough to be strong first candidates — intake, scheduling, document sorting, status updates — and rule out anything that touches privileged judgment, regulated advice, or final client deliverables at this stage. In the next 30 days, build and pilot a single workflow end to end with a human reviewing every agent action before it reaches a client, and use that pilot specifically to stress-test the audit trail and error-catching process rather than just the happy path. In the final 30 days, expand to a second workflow only after the first has run clean through a full billing cycle, and use that window to formalize the governance policy, the accountable owner, and the staff training adjustment so the third and fourth workflows can move faster without repeating the same groundwork. This sequencing matters more than the specific workflows chosen — firms that try to do governance and deployment simultaneously across multiple workflows tend to end up with neither done well.

Key Takeaways

• A genuinely capable agentic workforce is arriving in professional services faster than most firms' governance, training, and accountability structures are ready for it — treat that gap as the primary risk, not the technology itself.
• US professional services firms face this pressure from three directions at once: client expectations shaped by their own AI use, regulatory and liability exposure specific to licensed professional work, and visible regional competition on speed and pricing.
• Start with client-facing, repeatable workflows — intake, scheduling, document sorting — rather than judgment-heavy or regulated work, and keep a human reviewing every agent action until a workflow has run clean through a full cycle.
• Any client portal or payment flow an agent touches needs the same fraud and identity-verification discipline ecommerce businesses already apply, since faster automated flows create new exposure.
• Don't let marketing and visibility work pause while governance gets sorted — multi-location SEO and explainer video content for new service lines can run in parallel with the operational rollout.
• Budget in phases: prove the model on one workflow at the Essential tier before committing to a firm-wide Enterprise rollout.

The firms that get ahead of this won't be the ones that moved fastest — they'll be the ones that sequenced deployment against real governance, kept a human accountable at every client-facing step, and used the transition to actually reinvest in how their junior staff develop judgment rather than just cutting cost. If you want help mapping which of your firm's workflows are strong first candidates and what governance needs to exist before you deploy them, book a meeting with our team and we'll walk through it against your specific practice structure.

Frequently Asked Questions

What does "agentic workforce" mean in a professional services context?

It refers to AI agents that can independently plan and execute multi-step tasks — like client intake, document sorting, or first-pass research — rather than simply answering questions or offering suggestions inside a human's workflow. The key difference from a chatbot or copilot is that an agent completes a sequence of actions and only escalates to a human at defined decision points.

Is this the same thing as the chatbots my firm already uses?

No. A chatbot answers individual questions in a single exchange, while an agent carries out a multi-step process end to end — for example, qualifying a lead, checking calendar availability, and booking a consult with the right partner without a human doing each step manually. Most firms already have chatbot-level tools; the agentic shift is a step beyond that.

Why is this described as arriving "faster than firms are organizationally ready"?

Because the underlying AI capability has advanced quickly enough to handle real structured work, while the governance, accountability, and training structures inside most firms haven't been redesigned to match. The technology isn't waiting for firms to finish deciding who's accountable for an agent's output.

Who is Rishad Tobaccowala and why does his commentary matter here?

He's a longtime marketing and media strategist whose industry commentary in 2026 highlighted this specific gap between agentic AI capability and organizational readiness. This post treats that commentary as the trend signal grounding the discussion, without attaching invented statistics to it.

Does this mean professional services firms will replace staff with AI agents?

The more immediate risk isn't wholesale replacement — it's that firms without a deliberate plan will either automate too cautiously and lose ground on price and speed, or automate too fast without governance and expose themselves to errors reaching clients. Staffing changes, where they happen, should follow a redesigned workflow, not precede one.

Which types of professional services firms in the USA are most exposed to this shift?

Firms whose core work is structured, document-heavy, and rules-based — law firms handling contract review and discovery, accounting and tax practices, financial and management consultancies — have the largest share of tasks that current agentic systems can meaningfully assist with or execute.

What's the first workflow a firm should automate with agents?

Client-facing but low-judgment workflows: intake qualification, consultation scheduling, and document collection or sorting. These are repeatable, don't touch privileged advice or regulated deliverables, and let a firm build governance experience before moving to higher-stakes workflows.

What should NOT be handed to an agent yet?

Anything involving final legal advice, regulated financial recommendations, privileged judgment calls, or client deliverables that go out without a licensed professional's review. Agents are strongest on the structured groundwork feeding into that judgment, not the judgment itself.

How does this affect client trust in a professional services firm?

Client trust in these firms has traditionally rested on a licensed professional standing behind every deliverable. Introducing agents into the workflow doesn't remove that expectation — it raises the bar for showing clients exactly where a human reviewed and approved the work, which is why an audit trail matters as much as the automation itself.

Will clients notice or care if agents are doing some of the work?

Increasingly, yes. Clients who use AI internally are already asking pointed questions about turnaround time and pricing that assume some automation exists on the firm's side. Firms that can explain their approach clearly tend to fare better than firms that avoid the topic.

How should a firm price services differently once agents speed up delivery?

This is a live strategic question without a universal answer, but the general pattern is that firms relying purely on hourly billing for tasks an agent now completes in minutes will face pricing pressure from competitors who've restructured around value-based or flat-fee pricing for those same tasks.

What happens to junior associates and analysts if agents absorb entry-level tasks?

This is one of the harder organizational questions firms need to address directly rather than let resolve itself through attrition. If agents absorb the repetitive tasks that used to train junior staff, firms need a deliberate alternative path for developing the judgment those staff would otherwise have built on the job.

How long does it typically take to deploy a first agentic workflow?

A single, well-scoped workflow — such as client intake and scheduling — can typically move from mapping to a working pilot within 60 days when the underlying systems (calendar, CRM, document storage) are ready to be integrated. Firm-wide rollouts spanning multiple practice groups take substantially longer and should be phased deliberately.

What does an "Essential" tier engagement typically include?

At the Essential tier, the typical scope is a single agentic workflow for one practice area or function — for example, an intake and scheduling agent — starting around $1,000, without the multi-system integration or full compliance tooling a larger rollout would need.

What's different about a "Growth" tier engagement?

Growth-tier work typically connects multiple workflows — intake, portal communication, and internal document handling — with basic governance and audit logging built in, starting around $2,000, suited to a firm ready to move beyond a single pilot.

When does a firm need an "Enterprise" tier engagement?

Enterprise-tier work, starting around $4,000+, fits firms rolling out agentic workflows across multiple practice groups with integration into existing case or practice management systems and full audit and compliance controls — typically mid-size to larger firms with more complex systems to connect.

Does a bigger firm automatically need the Enterprise tier?

Not necessarily. Firm size matters less than how many systems the agent needs to touch and how much regulatory or compliance overhead applies. A smaller firm with complex compliance needs can outgrow the Essential tier quickly, while a larger firm piloting one workflow may start at Essential deliberately.

What ongoing costs should a firm expect after the initial build?

Beyond the initial build, expect ongoing costs tied to monitoring, periodic audits of agent decisions, and workflow adjustments as practice needs change — these should be scoped explicitly during the initial engagement rather than assumed to be zero after launch.

How does agentic automation affect a law firm's client intake process specifically?

An intake agent can qualify a prospective client against the firm's practice areas, check for conflicts against existing matter data, and schedule a consultation with the appropriate attorney — compressing a process that often takes days of back-and-forth into same-day scheduling, with a human reviewing before anything client-facing is finalized.

How does this affect accounting and tax practices during peak season?

Structured, repeatable intake tasks — client document collection, categorization, and status updates — are strong candidates for agentic handling during high-volume periods, freeing staff capacity for the judgment-heavy review and advisory work that peak season actually demands.

How does this affect management and strategy consultancies?

Consultancies often have heavy research synthesis and first-draft deliverable work that agents can accelerate, but the client-facing strategic recommendation itself should remain squarely a human judgment call, reviewed and owned by the engagement lead.

What is the biggest technical risk of deploying agents in a client portal?

The biggest risk is an agent taking an irreversible or client-visible action — sending a communication, processing a payment, finalizing a scheduling change — without a defined checkpoint, especially before the firm has built the audit trail needed to catch and correct an error quickly.

How does this connect to fraud risk on a client portal?

Once a portal accepts online retainer payments or deposits and an agent is involved in that intake or billing flow, it introduces exposure similar to what ecommerce businesses manage around chargebacks and account takeover — worth reviewing alongside standard fraud-prevention practices for online payment flows.

Should a professional services firm worry about chargebacks the way an online store does?

If the firm processes retainer or invoice payments online, yes, to a meaningful degree — a faster, agent-assisted intake and billing flow can reduce the manual friction that used to catch obvious anomalies, so the same disciplined verification practices used for ecommerce transactions are worth applying.

What governance should be in place before deploying any agent?

At minimum: a documented policy on which tasks an agent can complete unsupervised versus which require human sign-off, a full audit trail of agent actions, and one named accountable owner for the agent's performance — not a committee, and not left undefined.

Who inside a firm should own agentic AI governance?

Ownership should sit with someone who already has accountability for the underlying function — a practice group leader for practice-specific agents, or an operations or technology leader for firm-wide workflows — rather than being assigned informally or left ambiguous.

How does a firm audit what an agent actually did?

By maintaining a structured log of every action the agent took, the data it used, and the outcome, tied to the specific client matter or interaction — the same standard of traceability a firm would expect from a human staff member's work product, just captured automatically.

What compliance considerations apply to US law firms specifically?

State bar rules on AI-assisted work product, client confidentiality obligations, and malpractice liability standards all bear on how an agent can be used in matter-related work — firms should confirm their approach against their specific state bar guidance before deploying agents on anything client-facing in a legal matter.

What compliance considerations apply to US accounting and financial firms?

IRS and SEC scrutiny of automated processes touching client financial data means accounting and financial services firms need clear documentation of what an agent did and why, particularly for anything feeding into a filing, audit, or financial recommendation.

Can agentic AI handle regulated financial advice on its own?

No — regulated financial recommendations should remain a licensed professional's responsibility, with any agent-assisted research or drafting clearly reviewed and approved by that professional before it reaches a client, both for compliance and liability reasons.

How do I know if my firm is "organizationally ready" versus just "technically ready"?

Technical readiness means the tools exist and can be integrated. Organizational readiness means governance, accountability, staff training, and error-handling processes are defined before deployment — a firm can have excellent technical readiness and still be organizationally unready if nobody owns the agent's output.

What's a realistic first step if my firm hasn't started at all?

Map which client-facing and internal workflows are structured and repeatable enough to be strong candidates, rule out anything touching privileged judgment or regulated deliverables, and pilot exactly one workflow with full human review before expanding — resist the urge to move on several fronts at once.

How does marketing fit into this transition?

Marketing and client acquisition should keep running in parallel with the operational rollout rather than pausing for it — firms explaining new AI-assisted service lines are finding explainer video content converts well, and firms with multiple offices should make sure each location has a properly built local page.

Why would a professional services firm need AI video ads?

Short explainer videos are proving more effective than static text pages for building trust around a new service line or explaining how a firm uses AI responsibly — a message that's hard to convey convincingly in a paragraph of website copy alone.

Why does multi-location SEO matter for a firm with several offices?

Each office location competes in its own local search market — a firm with a Chicago and an Austin office, for example, needs a properly built local page for each city rather than one generic page, or it cedes local visibility to competitors who've done that work.

Does a single-office firm need to worry about multi-location SEO at all?

Not in the same way, though a growing single-office firm expanding into a second city should plan for a proper local page structure from the start rather than retrofitting it later once search visibility gaps have already formed.

How does agentic automation change turnaround time for clients?

For structured tasks like document intake, initial research synthesis, or status updates, turnaround can shrink from days to hours or less, since the agent doesn't wait on staff availability to start the work — though final review and judgment steps still take the time they need.

Will faster turnaround from agents hurt or help client retention?

It tends to help, provided quality and accountability are visibly maintained — clients generally value a faster, transparent process, but a mistake reaching a client because a review checkpoint was skipped will damage trust faster than slow service ever would.

How should a firm measure whether an agentic workflow pilot is working?

Track error rate against the audit log, time saved versus the manual process it replaced, and whether the workflow completed a full billing cycle without an issue reaching a client — expansion to additional workflows should wait until those measures are solid.

What's the risk of moving too slowly on this trend?

Firms that wait too long risk losing ground to competitors offering faster turnaround and more competitive pricing on routine matters, and they lose the runway to build governance and staff training deliberately instead of reactively once a competitor's move forces their hand.

What's the risk of moving too fast without governance?

Deploying agents into client-facing or judgment-heavy work without a defined review process and audit trail risks an error reaching a client before anyone catches it, which is a more damaging outcome for a professional services firm than moving a few months slower with proper controls in place.

Can smaller professional services firms compete with larger firms on this?

Yes — smaller firms can often move faster precisely because they have fewer approval layers, provided they still take governance seriously. A well-scoped Essential-tier pilot can let a smaller firm match a larger competitor's turnaround on specific routine matters without matching their headcount.

How does this trend affect firms that serve clients across multiple US states?

Firms operating across state lines need to account for varying state bar or licensing rules on AI-assisted work product in each jurisdiction, which makes a documented, jurisdiction-aware governance policy more important than for a single-state practice.

What role does client communication play in this transition?

Being upfront with clients about how and where AI assists in their matter — without overstating or understating it — tends to build more trust than staying silent on the topic, especially as clients increasingly expect some level of automation from their service providers.

How do I integrate an agent with our existing practice management or CRM system?

Integration scope depends on the specific system in use — most modern practice management and CRM platforms expose APIs that an agentic workflow can connect to for scheduling, document handling, and status updates, which is typically scoped and confirmed during the initial workflow-mapping phase of a build.

What happens if an agent makes a mistake in a client-facing action?

This is exactly why a defined review checkpoint and audit trail matter before deployment — the goal is to catch and correct an error before it reaches the client, and to have a clear record of what happened and why if it doesn't.

Is this trend specific to the USA, or is it global?

The organizational readiness gap Tobaccowala's commentary describes is a global pattern, but the specific pressures explored in this post — US regulatory exposure, state-by-state licensing rules, and the visibility of regional competition among US firms — are shaped by the US market context.

How quickly is this expected to change over the next year?

A precise timeline isn't publicly available, and this post won't invent one — the safer planning assumption is that agentic capability will keep improving steadily, which argues for building governance and piloting workflows now rather than waiting for a clearer signal that may not come before competitors have already moved.

What should a firm avoid doing in response to this trend?

Avoid two extremes: freezing all AI adoption out of caution while competitors move, and deploying agents broadly across client-facing and judgment-heavy work without governance in place. The middle path — a scoped pilot with real oversight — is the one that holds up over time.

Where should a professional services firm start if it wants help with this?

Start by mapping which of the firm's own workflows are structured and repeatable enough for a first pilot, then scope a single agentic workflow with proper governance built in from day one — booking a meeting with a team experienced in AI agent builds for professional services is a practical way to get that mapping done accurately.