The Data Gap
Published November 10, 2026 · By The Crossing Report · 21 min read
This issue: 78% of corporate clients say receiving AI-quality improvements from their professional services firm is "very important" or "essential." Only 6% say most or all of their providers are actually delivering. This issue names the gap — and the specific data connections that close it for your firm type.
Your Clients Are Already Grading You on AI
Thomson Reuters just published its 2026 AI in Professional Services report — 1,816 professionals surveyed across legal, tax, accounting, and consulting. Two numbers belong together.
78% of corporate clients buying professional services say receiving AI-enabled quality improvements is "very important" or "essential." That is not a projected number. It is reported by clients who are buying professional services right now.
6% — the share of firms that most or all of those clients say are actually delivering on that expectation.
The 72-point gap is not a technology gap. It is a data connection gap. The firms that are delivering are not necessarily using better AI tools. They are using AI tools that can see the same data their professionals see: client relationships, matter history, candidate records, engagement methodology. The 94% that are not delivering are running AI on top of data silos — general-purpose tools pointed at generic prompts, disconnected from the firm's proprietary intelligence.
Three general counsel — from Fifth Third Bank, Bayer, and Progressive Insurance — named it directly in a Law360 panel last month: they now expect outside counsel to proactively disclose AI use and explain how it improves the work. Not wait to be asked. Raise it. Most firms have not had that conversation with a single client.
Your clients are grading you. The scorecard is already running. Whether you know what is on it or not.
The rest of this issue is about closing the gap — by firm type, with the specific tools that now exist to connect your proprietary data to AI.
The Staffing Firms Growing 3–5x Have One Thing in Common
Bullhorn's 2026 GRID Industry Trends Report names the single variable separating the staffing agencies gaining ground from the ones staying flat.
Not which AI tool they use. Not firm size. Not market specialization.
Whether AI is embedded throughout the ATS — or bolted on outside it.
3–5x revenue growth for agencies with AI embedded throughout their workflow versus agencies running general-purpose tools on top of their core systems. 54% of agencies with AI-assisted screening report KPI gains above 25%. But only 10% have reached full workflow integration. 90% of the sector is still running AI as a surface layer over the same siloed data infrastructure.
The mechanics are now documented. At fully integrated agencies, the recruiter's day has inverted: work that used to be 80% administrative and 20% relationship-building is now the reverse. AI handles sourcing runs, structured screens, scheduling (8 seconds versus the multi-email manual sequence), and CRM updates. The recruiter manages exceptions, handles client relationships, and closes. The 57% worker redeployment rate at fully embedded agencies — placing more candidates from the existing database at lower marginal sourcing cost — is the revenue multiplier.
The entry question for a staffing firm owner this week: for each AI tool currently running, does it live inside your ATS or outside it? The outside-ATS list is your integration roadmap.
Harvey and Legora Launched the Same Product on the Same Day. That's Not a Coincidence.
On October 13 and 14, two of the best-funded legal AI companies in the world each announced an intake product for law firms. Not in the same quarter. In the same 48-hour window.
Harvey launched Harvey Intake. Legora launched Legora Intake. Both systems connect a firm's front door — first client contact, matter intake, triage — to AI. Both were developed independently. Both shipped simultaneously.
When two independent vendors find the same white space at the same moment, that is a market signal: intake and triage is the next major AI category in legal, following research and document drafting. After AI replaced the second draft, it is moving to the first conversation.
For a 5–20 attorney firm, intake is the workflow most consumed by paralegal and admin time. A client calls, staff takes notes, someone manually enters data into the matter management system, someone checks conflicts, someone prepares the intake summary. That workflow has existed essentially unchanged for thirty years.
One calibration: the Legal Stack 2026 report on firms under 20 attorneys puts true AI adoption — regular, workflow-integrated use — at 28–35%, not the 71–75% usually cited. The higher figure counts occasional experimentation. Most small law firms are closer to the beginning of this than the headlines suggest.
If you are already on Clio: Clio Duo AI drafting is included in your Clio Manage subscription at no marginal cost. Before evaluating Harvey or Legora Intake, check whether it is activated on your account.
Clio's 2026 Legal Trends Report — released at ClioCon 2026 on October 26 — puts a number on the shift: lawyer utilization rates are projected to hit 42% of the workday billable in 2026, the largest annual jump in the report's 11-year history. 85% of legal professionals report AI adoption at their firms. Firms running Clio Work and Manage together record 10% higher revenue per responsible lawyer and 6.5% more billable hours. These are your benchmarks. If your utilization and revenue per lawyer are not moving in this range, your AI tools are installed but not integrated — the same data-connection gap this issue describes.
What the Market's Answer Looks Like
The 78%/6% gap is real. So is the answer.
On October 22, Thomson Reuters launched CoCounsel Tax, Audit & Accounting — a fully rebuilt agentic AI platform for accounting and tax professionals. It does not suggest. It builds: amortization schedules, spreadsheets, tax memos, multi-tab Excel workbooks with formulas, first-pass document analysis. The work is delivered for human review and sign-off before anything goes to a client or files. Grounded in Checkpoint — 12 million primary source documents maintained by hundreds of CPAs and tax attorneys — not a general-purpose language model trained on the public internet.
Thomson Reuters calls it "Fiduciary-Grade AI." The distinction matters: when your professional signature goes on work, the AI assisting with it should be held to a professional standard. TR stock moved 7% on announcement day.
The AICPA CEO said it directly this week: "AI is not going to disrupt the accounting profession, but it will change what an accountant does." Bureau of Labor Statistics projects 5% employment growth for accountants through 2034 — faster than average. The work AI handles best (rules-based, repeatable, structured) competes on price. The work AI handles poorly (judgment, negotiation, regulatory interpretation, complex client relationships) commands premium fees. CPA firms with proprietary client data and deep advisory relationships are not being replaced. CPA firms doing commodity work without AI are.
CoCounsel is the market's answer to the data gap for accounting. The 78% of clients who expect AI-quality delivery now have a specific platform to point to — one their CPA can actually use.
One week before CoCounsel's launch, Wolters Kluwer launched agentic tax return preparation inside CCH Axcess Tax — the platform used by tens of thousands of U.S. accounting firms. The structure is identical: the system classifies source documents, maps data to the appropriate tax forms, executes calculations, and flags exceptions before the preparer opens the file. On October 26, Wolters Kluwer expanded the launch to include agentic audit capabilities. Two of the three largest U.S. accounting platforms launched agentic tax preparation in the same week. If you are on CCH Axcess, the capability is already inside your platform — no additional purchase required. If you are evaluating platforms, the gap on this specific feature has closed.
What Intuit Just Did for Your Firm (and What to Read Between the Lines)
Intuit Connect wrapped three days of announcements in Las Vegas. Here is what matters for a firm owner who was not there.
The QB Live pivot. QuickBooks Live is no longer actively marketed to accountant-attached businesses. Intuit is repositioning around the ProPartner program — five tiers replacing the old ProAdvisor structure, with a 2027 transition timeline. The honest read: Intuit is stepping back from direct-to-consumer bookkeeping for clients who already have an accountant. This is not a structural concession — the competitive tension between Intuit and the accounting profession has existed for years and is not gone. But for the next 12–18 months, the framing works in your favor. Intuit wants your firm to be the advisory layer on top of their platform.
The cross-client view. Intuit Intelligence is in beta for IAS Accelerate subscribers ($149/month). The headline capability: a portfolio-level AI view across all your client files simultaneously. Instead of opening individual files, ask: "Which clients have deteriorating cash flow this month?" and get an answer drawn from your actual portfolio. For an advisory firm with 15 or more clients, that cross-client intelligence is materially different from anything available before. For a pure compliance shop, the math on $149/month depends on whether you are building advisory capacity.
Three new agents. Payments Agent, Accounting Agent, and Project Management Agent confirmed at Connect. Intuit's stated time savings: 12 hours per month. Calibrate that against your actual workflow before accepting it. All three follow the draft-for-review model — they flag and propose; a human approves before execution. 88% of accountants say they use AI; only 6% trust autonomous execution (CPA Trendlines/McCarroll data). The agents are built for the 88%, not the 6%. Human-in-the-loop is not a limitation — it is the correct implementation model for licensed professionals.
Day 2 announcement worth tracking. Intuit launched Agentic Book Close — accountants upload their own SOPs and the AI builds automated workflows from them: reconciliation, balance sheet review, financial summaries run according to the firm's documented process, not a vendor template. The distinction matters. Every other AI bookkeeping tool applies its own logic to your data. Agentic Book Close applies your logic. The prerequisite: your close process has to be documented. For firms with written checklists, this is immediately applicable. For firms without them, documenting your process is now Step 1 — and that documentation is valuable IP regardless of which AI tool you ultimately use.
Agent Studio. The capability Intuit spent the least time on may matter most to a 10-person firm. Agent Studio lets you encode your firm's own workflows and standards into custom AI agents — in plain language, without writing code. Not just the monthly close: any process the firm runs repeatedly. How you onboard a new client. How you review a payroll run. The standards you apply before anything leaves the firm. You describe the process in natural language; Agent Studio turns it into a deployable agent that runs it consistently. The distinction from every other AI tool is institutional: the agent follows your judgment, not a vendor's template. The prerequisite is always the same — the process has to be written down first. Start there.
Firm Playbooks (IAS Accelerate, available in beta). If Agentic Book Close is Step 2, Firm Playbooks is Step 1. The feature — available in IAS Accelerate since August — lets an accounting firm build one standardized template per client type: chart of accounts preferences, accounting policies, company settings, engagement workflow steps. Build it once for construction clients; apply it instantly to every construction client you onboard. The practical two-step path: build your Playbook templates for each client type, then upload those same SOPs into Agentic Book Close. The AI executes your documented process at scale — not a vendor template — every time.
Intuit also announced a multi-year partnership with Anthropic to build custom AI agents that combine QuickBooks financial data with Claude's reasoning. The beta is gated to Intuit Enterprise Suite (mid-market, custom-quoted) — not an immediate action item for a 10-person firm. But the direction is clear: the combination of Intuit's financial data graph with Claude's reasoning is what makes "ask your accounting software anything and get a genuinely useful answer" technically feasible. Firms building toward this should audit their data quality inside QuickBooks now. The firms that will benefit most are the ones whose data is cleanest.
Three evaluation questions for any Intuit Connect feature: (1) Does it require encoding your firm's methodology into a third-party platform you do not own? (2) Does the claimed time savings reflect your actual workflow, not the demo workflow? (3) Is human oversight built in, or does it need to be configured?
Two urgent sidebar items the Connect keynote did not emphasize. QBOA retires December 31. Every accounting firm on QuickBooks Online Accountant must migrate to the new Intuit Accountant Suite before year-end — Core is free; Accelerate is $149/month. Separately: QuickBooks raised prices 70% on QBO Advanced on August 1. If you have not repriced client engagements to reflect the new $340/month QBO Advanced tier, that margin conversation is overdue.
Amelio AI offers a different model: your firm builds its methodology first, and AI executes within the logic you defined — not the other way around. Worth evaluating for 10–20 person CPA firms with established processes who are not ready to embed their methodology in QuickBooks specifically.
Two acquisitions signal where the market is heading: CohnReznick acquired Aldrich (Top 100 CPA firm), and Citrin Cooperman acquired Rosetree Solutions (an AI consultancy). The consolidation map is filling in. Independent CPA firms have three options: merge up, specialize deeply, or build AI capability fast. The window on doing nothing is closing.
Your Practice Data Is Now Queryable
In October, MCP — the protocol that connects AI assistants directly to business software — landed in both accounting and law simultaneously.
For accounting firms: Digits Write MCP (October 13) lets AI propose general ledger corrections — recategorizations, journal entries, bill fixes — with a human approval queue before anything posts. Financial Cents MCP (October 20) connects your practice management system to Claude, ChatGPT, and Gemini: ask "Which invoices are unpaid this month?" and get an answer drawn from your actual client list. Dext MCP (October 22) adds client financial health scores, live cash position, and time-consumed data to the same interface. Three tools in two weeks. The MCP stack for a small accounting firm is now complete. Your practice management talks to Claude. Your general ledger talks to Claude. Your accounts receivable data talks to Claude. That is the full picture.
For law firms: MyCase MCP (73 supported actions across cases, clients, documents, billing, and intake) gives Claude direct access to case management data — ask "Which matters have open invoices past 60 days?" and get a live answer from your actual matters. Harvey's Connector Library (generally available, October 21) adds document write-back to iManage, NetDocuments, SharePoint, Google Drive, and DocuSign — work completed in Harvey pushes directly back into your document management system, removing the copy-paste step that has been the single largest AI workflow friction point in legal.
Among small law firms, Claude holds 40% share in the CosmoLex 2026 legal tech survey — the largest AI platform among respondents surveyed. The top concern (accuracy, ranked above security and compliance) is the discipline that matters most: verify before every client-facing output.
For firms already on Clio: Vincent's agentic mode (launched April 2026, now in Clio Work) changes what delegation looks like. Instead of prompting once and reviewing output, you state a multi-step goal — "draft a demand letter, log it in the matter, and schedule a follow-up" — and Vincent executes the sequence without re-prompting. The distinction between "AI assistant" and "AI agent" is this: an assistant responds to each prompt; an agent executes toward a goal. For a solo or 5-attorney firm, agentic Clio means genuinely delegating multi-step workflows, not just faster search.
On October 28, Anthropic added six ready-made Legal Artifacts to Claude — interactive templates a small firm attorney can open inside Claude and use the same day, with no IT integration or vendor setup: Citation Checking, Contract Review, Matter Tracking, Regulatory Monitoring, Jurisdiction Comparison, and Multi-Contract Review. For firms preparing for SB 574's attorney verification obligation (effective January 1, 2027), the Citation Checking artifact surfaces mismatched quotations, incorrect page references, and unsupported claims in filings and memos before they go out. Contract Review flags deviations from your own standard terms — not a generic playbook, your terms. Search "Legal Artifacts" inside Claude; the gallery is there on a standard Professional or Team plan. No new vendor, no contract, no IT ticket.
One forward signal: August AI lets law firms build productized services from existing expertise — a standard lease review, an NDA package, an employment agreement offering — delivered by AI working from the firm's own precedents, with lawyer review before delivery. The economics: expertise codified once, sold at a fixed fee repeatedly. Early-stage, but this is the model that converts hourly income into recurring revenue.
The Compliance Task 70% of Tax Firms Haven't Done Yet
Deadline: December 31, 2026.
IRS OPR Alert 2026-19 requires every Circular 230 practitioner to update their Written Information Security Plan to include AI tools before PTIN renewal. Specific obligation: name the AI tools your firm has approved for use, specify who may use them, and define what client data cannot be entered into any AI system.
If your firm has been using ChatGPT, Claude, Copilot, or AI-embedded features in QuickBooks or your tax prep software, that use must be covered in your WISP before you renew. The FTC Safeguards Rule adds a parallel obligation: when new technologies enter your workflows, your security plan must reflect them.
I updated our internal security plan last month. It took 40 minutes. Most of what I added were three AI tools we had been using all filing season — tools that had never appeared in the plan because we added them incrementally and never circled back to update the paperwork.
Five lines to add to any small firm WISP: (1) AI tools approved for firm use and which staff may use them; (2) client data classification — what cannot be entered into any AI system; (3) output review requirements before client delivery; (4) incident response for AI-related data exposure; (5) a review date for this section — set it quarterly, not annually.
The Firm That Replaced 11 Staff With AI Agents
Aicila Group, a recruitment firm, replaced a team of 11 with AI agents handling the full recruitment workflow: business development, candidate sourcing, client communication, follow-up, nurturing. This is a documented case of complete workflow-level human-to-AI staff replacement — not a pilot, not an experiment.
This is not a prescription. But the case is useful because it makes the frame concrete: the roles that were replaced were high-volume, structured, repeatable, and contained few genuine judgment calls.
The question for your firm is not "should I replace my team?" It is: "Which pieces of my delivery model have the same characteristics — high-volume, structured, repeatable interactions where judgment calls are less than 20% of the work?"
The sector-wide data is clear. Enterprise agent deployment in staffing jumped from 11% to 42% in six months. The 80/20 flip — from 80% administrative to 80% relationship-focused workdays — is the documented norm at AI-embedded agencies. Staffing Industry Analysts' benchmark is at 106 for 2026, up 9% year over year. Hours are up, but AI skills demand has grown 282% since May 2024. The agencies capturing that demand have embedded AI and an updated candidate bench.
For staffing firms operating in Connecticut: The WARN Act provision is live now. Any workforce reduction notice involving AI-driven changes requires explicit disclosure language in the filing. The full candidate-facing disclosure requirement (written notice to applicants interacting with AI screening) takes effect October 1, 2027. Wire it into new workflows now rather than retrofitting next fall. California, New York, and New Jersey have active bills modeled on Connecticut's approach.
The Retainer Isn't Being Cancelled. It's Being Cut in Half.
60% of senior US marketing leaders report spending less on agencies because of AI. Not canceling the contract. Reducing the retainer.
The pattern is specific: a $25,000/month retainer becomes $15,000/month as clients start producing AI-assisted content, basic analytics, and social scheduling in-house. Budget pressure overtook performance as the top reason clients leave agencies in 2026. 41% of agencies face measurement challenges as traditional attribution models lose reliability.
Webflow Campaigns — launched last month — illustrates exactly what clients are discovering they can do themselves: it extracts a brand's design system from an existing Webflow site, generates a campaign brief, and builds landing pages and ad creative autonomously, with Claude MCP integration for natural-language modifications. That is the work that used to live on a retainer.
I ran the AI tool cost audit on our lead generation agency this year. AI tool spend went from $1,200/month to $3,400/month in twelve months — three tools had doubled or tripled in cost without a single retainer conversation. HubSpot's shift to outcome-based pricing on Breeze, Zapier's model-tier migration — these are not isolated events. The platforms are repricing their value upward. If your client retainers have not moved, your margin is being compressed quietly.
The agencies repositioning are not competing for production work. They are charging $1,500–$8,000/month for AI strategy consulting: teaching clients to build with AI themselves, managing the strategic layer, handling the work AI cannot produce without judgment. That is not a smaller business. It is a different business, with better margins. Deloitte's 2025 benchmark puts it in numbers: value-based firms grew annual revenue at 8.7%; hourly firms at 2.1%. The pricing model is now a performance variable, not just a preference.
This Week: One Data Connection
The goal this week is not a new tool. It is one data connection — the specific place where your AI and your proprietary data are not talking.
Law firms: CosmoLex's 2026 Legal Tech Report found that 86% of small law firms have not adjusted their pricing since adopting AI. If you are one of them, that is a concrete problem: AI tool costs are rising, billable hours are compressing, and IRS Circular 230 guidance now creates billing risk if AI-assisted time savings are not reflected in fees. One action this week: look at one recurring matter type and determine whether a flat fee or retainer model would better reflect the actual delivery cost post-AI. That is also the conversation to have before a client brings it up.
Accounting firms: If your firm is on QuickBooks, activate the Intuit Intelligence Unified Inbox this week — it is live, included in your subscription, and it is the "data connected to AI" moment in practice. Success criterion: one client file reviewed through the AI portfolio summary without manually opening three separate tabs.
Staffing firms: For each AI tool you are currently using, answer one question: does it live inside your ATS, or does it require your staff to leave the ATS to use it? List them. The outside-ATS column is your integration roadmap. Before building any outside-ATS integration, call your ATS vendor and ask what is coming natively — do not build connections that will be replaced in six months.
Consulting and marketing agencies: Pull your AI tool subscriptions for the last six months. What did each tool cost in Month 1 versus today? Identify one tool where the cost increased without a corresponding engagement adjustment. That is your Q1 pricing conversation. Worth having before the client brings it up.
Issue #30 | The Crossing Report | November 10, 2026
82% of professional services firms have adopted AI. Only 18% have any way to measure what it's delivering. Next issue: the 4-metric scorecard that tells you in two hours whether your AI investment is working — and which workflows to prioritize first.
This is the kind of intelligence premium subscribers get every week.
Deep analysis, cross-sector patterns, and the frameworks that help professional services firms make the crossing.