AI Training at Professional Services Firms: Your Staff Wants to Learn — But You Haven't Created the Path

Here is the counterintuitive finding that should change how you think about AI at your firm: the people most motivated to learn AI are the ones you're least likely to train.

The Federal Reserve Bank of New York published research in April 2026 documenting this exactly. Workers without college degrees — the support staff, billing coordinators, administrative assistants, junior bookkeepers, and entry-level paralegals at your firm — place the highest value on AI training of any worker segment. They are eager. They see what AI can do. They want to learn.

And they are receiving the least employer-provided training of anyone in the workforce.

Meanwhile, 59% of enterprises report an "AI skills gap" even after providing some training. Only 44% of US workers have received any AI training at all. Just 16% receive it regularly.

At a 10 to 40-person professional services firm, that statistic maps directly onto a risk you are running right now.

---

What the Federal Reserve Research Found

On April 14, 2026, the Federal Reserve Bank of New York published "Use of Gen AI in the Workplace and the Value of Access to Training" through its Liberty Street Economics blog.

This is not a vendor report. This is Federal Reserve Bank research — citeable, credible, and methodologically rigorous.

The central finding: workers without college degrees place the highest subjective value on AI training, but have the lowest rate of receiving it from their employers.

The data points:

• 65% of US workers want AI training
• Only 44% have received any AI training
• Only 16% receive it "often"
• Workers without degrees value training more than college-educated workers — and receive it far less
• HR Dive summarized the downstream effect bluntly: "AI use is happening in silence amid lack of training"

That last line matters. When you don't train staff, they don't stop using AI. They use it anyway — on their own terms, with their own tools, and without any guardrails your firm set.

---

Who This Is at Your Firm

In a 10 to 40-person professional services firm, the workers the NY Fed is describing are your:

• Billing coordinators and accounts receivable staff — handling invoices, following up on collections, tracking client payments
• Administrative assistants — scheduling, client communications, document management
• Junior bookkeepers and staff accountants — data entry, reconciliations, routine reporting
• Entry-level paralegals — document review, drafting standard correspondence, research pulls
• Scheduling and intake coordinators at staffing or consulting firms — first-contact work, screening, logistics

These are your highest-ROI AI automation targets. The work is routine, high-volume, and repeatable. AI can cut the time-per-task dramatically.

And these staff members are motivated. The NY Fed data shows they are not waiting for permission to care about AI — they are already placing a high value on learning it. The barrier is not aptitude. The barrier is access.

What happens without access? They improvise. They use personal Claude or ChatGPT accounts. They email client documents to a free tool. They build workflows on platforms that have no data agreements with your firm. This is already happening. DataCamp's 2026 research on AI training gaps confirms it: when formal training is absent, informal and unsanctioned adoption fills the vacuum.

That is a liability problem. It is also a retention problem, which we'll get to next.

---

The Two Risks You're Running If You Don't Act

Risk 1: Shadow AI Exposure

Staff who want to use AI but haven't been given a sanctioned tool and policy will use whatever is available. For most support-level staff, that means personal accounts on ChatGPT or Claude with no data handling agreements, no audit trails, and no firm oversight.

A billing coordinator pasting client invoices into a free AI tool. A junior paralegal running a deposition summary through a browser-based AI that logs prompts. An administrative assistant using a personal account to draft client emails.

Your client confidentiality obligations do not pause because your staff was improvising. Your engagement letters do not include exceptions for unsanctioned AI tools.

For a detailed breakdown of this exposure and what a basic AI policy looks like, read: Your Staff Is Using AI on Client Work Right Now — and Your Firm Has No Policy.

Risk 2: Retention Pressure

The 2026 RSM Mid-Market Report and Robert Half's hiring data both point to the same trend: AI-enabled workers are becoming a competitive differentiator for employers, not just for clients. Staff who want to grow in AI but can't do it at your firm are actively considering firms that will provide that path.

This is especially acute for junior staff and support roles — the exact workers the NY Fed identifies as most motivated to learn. They're not passive. They want to build skills that make them more valuable. If your firm is not the place where that happens, they will find one that is.

The math on replacement costs for a billing coordinator or junior paralegal is not trivial: recruiting fees, onboarding time, productivity ramp, and institutional knowledge lost. Training one person for 30 minutes a week is not the expensive option.

---

What a Firm-Level AI Training Program Actually Looks Like

For a 10 to 40-person professional services firm, an AI training program does not require a budget line, a dedicated HR function, or a technology committee. It requires a decision, a tool, and one person willing to own it.

Here is what this looks like in practice:

Step 1: Pick one tool. Do not let "which AI tool" become a six-month evaluation project. If you're already on Microsoft 365, start with Copilot — it's already in your subscription. If you're not, Claude Pro ($20/month per seat) is the most capable general-purpose option for drafting, summarizing, and structured thinking tasks. For legal firms, CoCounsel or Casetext if budget allows; for accounting, explore what's built into your practice management software.

One tool. Firm-wide. Start there.

Step 2: Map three tasks per role. Have each department identify the three tasks they do most often that are routine, time-consuming, and document-heavy. For billing staff: invoice drafting, payment follow-up emails, collections summaries. For paralegals: deposition prep, client correspondence, standard research summaries. For admin: meeting notes, scheduling communications, intake forms.

These are your starting workflows. AI does not replace judgment — it accelerates the paper-pushing that surrounds it.

Step 3: Name one AI lead per department. Not a tech expert. Just a willing adopter — someone who is curious, will try things, and can share what works. Give them explicit permission to spend 30 minutes a week experimenting. Their job is not to become an AI expert; it is to discover what works for your specific workflows and bring it back to the team.

This is how AI actually spreads at small firms: person by person, workflow by workflow, not through company-wide mandates.

Step 4: Run a weekly 30-minute debrief. A Slack thread works. A standing team huddle works. The format is simple: what did you try with AI this week? What worked? What didn't? What are we doing next week?

This surfaces good workflows, kills bad ones quickly, and builds a shared vocabulary around AI use at your firm. After 60 days, you will have the foundation of a firm-specific AI playbook — which doubles as your onboarding manual for future hires.

---

The White House Context: This Is Now a Workforce Issue

AI training is not just a productivity conversation anymore. It is a policy conversation.

In January 2026, the White House AI Executive Order mandated that federal agencies build AI training pathways for workers — an explicit acknowledgment that AI capability is now a workforce development issue, not a technology issue.

On the legislative side, the Cantwell-Moran Small Business AI Training Act — bipartisan legislation reintroduced in 2026 — would create grants for small firms specifically to access AI education. For a full breakdown of what that bill covers and what it means for professional services owners, read: What the Cantwell-Moran Bill Means for Professional Services Firms.

The policy signal is clear: Congress and the executive branch both recognize the small business AI training gap. That bill may or may not pass this session. Your competitors are not waiting on Congress. Neither should you.

The practical takeaway: AI training is moving from competitive advantage to table stakes. Firms that build the training habit now will have a material head start over firms that wait for federal programs, vendor mandates, or industry associations to tell them to start.

---

Frequently Asked Questions

Do I need to spend money on AI training for my staff?

You need to spend time more than money. Most AI tools your staff needs are already available in existing subscriptions. The real investment is internal: one hour per week of structured practice and a designated person who owns adoption.

What if my staff uses AI wrong on client work?

That is almost certainly what is happening now, without training. A simple policy, a designated tool, and a 30-minute training session dramatically reduces risk compared to the unmanaged alternative — which is staff using personal accounts with no data hygiene and no audit trail. See the shadow AI post linked above.

Aren't AI tools too complicated for non-technical staff?

The NY Fed data directly challenges this assumption. Workers without college degrees — who typically hold support and junior roles at professional services firms — place the highest value on AI training. They are motivated. The barrier is access, not aptitude.

How do I know if my training is working?

Pick one measurable output: time to complete a specific task, number of drafts before final, client response time. Track it before and after 30 days. Thomson Reuters found only 18% of professional services firms measure AI ROI — making that measurement is itself a competitive edge.

---

Your Next Step This Week

Here is the one thing worth doing before this window closes:

1. Pick the tool — Copilot, Claude Pro, or one sector-specific option that is already in your stack
2. Name one person per department who will own adoption for the next 30 days
3. Block 30 minutes per week for four weeks — a team thread, a short huddle, whatever fits your rhythm

That is the entire program. Start there. Everything else follows.

If you want weekly signals on what is actually working at firms your size — not theory, not vendor hype, field reports from firm owners going through the same transition you are — subscribe to The Crossing Report.

One issue. Every Monday. Written for the owner, not the enterprise.