All case studies

Case studies: accounting firms

Stop keying invoices and K-1s by hand.

We automate AP and tax-document intake inside your existing stack: AI workflows that read invoices, extract tax-document data, and pre-draft reconciliations inside QuickBooks, Xero, your ERP, or a portal with no API. Your staff review and approve. Published results: 50% faster AP processing at one CA firm, and 75% less time on statement analysis in a peer-reviewed study.
  • −75%

    Time on income-statement analysis

    Peer-reviewed study

  • −50%

    AP processing time

    Mid-sized CA firm

  • ≈0%

    AP error rate after automation

    Mid-sized CA firm

  • 6-12 wks

    Typical first deployment

    ATI Lab delivery

Case studies

Firms that already automated AP, tax docs and analysis

Each case shows the bottleneck, the fix, and the published number. Sources are linked.

Third-party results from other companies, linked to source. They are not ATI Lab clients. How we would build it for you is below. See ATI Lab's own client builds.

Mid-sized CA Firm - AP Automation

Case study compositeThird-party result

The bottleneck

AP was the team's whole day: keying invoices, matching transactions, chasing mismatches.

What fixed it

AI-powered accounts payable processing deployed to automate invoice reading, data extraction, validation, and reconciliation.

Published result

  • −50%(decrease)

    processing time

  • ≈0%

    error rate (near-zero)

  • Advisory

    team freed to focus on client advisory vs paperwork

How we would build this AP workflow

  1. Invoice in

    Email, upload or portal

  2. Read + extract

    Vendor, amounts, lines

  3. Validate vs vendor

    Record and PO match

    Branch: Mismatch → flagged for review

  4. Draft reconciliation

  5. Staff approves

    Human approval
  6. Posted to ledger

IllustrativeHow we'd build it; the source doesn't detail approval steps.

AP processing time, indexed (before = 100)

  • Processing time50% less
    BeforeBefore: 100
    AfterAfter: 50
Indexed from the published 50% reduction. Error rate after automation: near zero. Third-party published result.

Source: AI Accountant (opens source in a new tab) - "AI Accountant Case Studies That Transformed CA Firms" (Jun 2025), partner firm testimonial

University of Aveiro - AI Agent + No-Code Study

Academic research - PortugalThird-party result

The bottleneck

Manual vertical analysis of income statements is slow, variable, and doesn't scale. GPT alone helped but still required significant human intervention.

What fixed it

AI agent combining GPT with no-code automation tools. 150 experimental analyses compared manual, GPT-only, and AI agent approaches using Design Science Research methodology.

Published result

  • −75%(decrease)

    execution time vs manual

  • More stable

    performance with less variability

  • Minimal

    human intervention needed

Execution time per analysis, indexed (manual = 100)

  • Manual vs AI agent75% less
    BeforeManual: 100
    AfterAI agent: 25
150 experimental analyses, peer-reviewed (Nov 2025). A GPT-only setup was tested too, but it needed heavy human input and the study gives no comparable figure, so it isn't plotted. Third-party published result.

Source: MDPI Digital (opens source in a new tab) - "AI Agents and No-Code Tools in Accounting: A Case Study" by Miguel Resende (Nov 2025), peer-reviewed

RSM + Additive AI Platform

Top 10 firm - USAThird-party result

The bottleneck

Complex tax documents like K-1s and partnership compliance packages required extensive manual processing. Time-intensive, error-prone, and hard to scale during tax season.

What fixed it

Partnered with Additive (Nov 2024) to deploy their GenAI platform for automated processing of complex tax documents, streamlining collaboration between tax professionals.

Published result

  • K-1 intake automated

    complex tax documents processed by the platform

  • Less manual keying

    with accuracy reported higher

  • Shared tax-team workflow

    collaboration between tax professionals

The source reports no figure for this case, so the results are shown as text.

Source: Firm of the Future / Intuit (opens source in a new tab) - "AI In Accounting" (Feb 2025)

Generations Wealth CPA (GWCPA)

Mid-size firm - Maryland, USAThird-party result

The bottleneck

Audit risk assessment and tax research consuming disproportionate staff time. Clients needed faster access to preliminary insights before advisory meetings.

What fixed it

Integrated MindBridge AI for audit risk assessment, Ask Blue J for tax research, and launched a custom GPT ("The Generations Advisor") in Dec 2024 for client self-service exploration of financial questions.

Published result

  • Multiple tools

    integrated across audit, tax, and client service

  • 24/7

    client access to preliminary financial insights

  • Shorter prep

    advisory meetings with pre-informed clients

Source: Firm of the Future / Intuit (opens source in a new tab) - "AI In Accounting" (Feb 2025)

CLA (CliftonLarsonAllen) - Engine B Acquisition

Context: how the top 10 is investingEnterprise referenceThird-party result

The bottleneck

Needed proprietary AI technology to offer differentiated services at scale. Off-the-shelf tools weren't enough.

What fixed it

Acquired UK-based tech firm Engine B to build their own AI technology stack. CEO Jen Leary positioned the move as "becoming a force for positively disrupting the profession."

Published result

  • In-house AI

    proprietary capability built through an acquisition

  • New services

    offered to clients

Shown for context. A small firm gets the same capability by automating one workflow, not by buying a tech company.

Source: Firm of the Future / Intuit (opens source in a new tab) - citing AICPA 2024 report on GenAI for small accounting firms

How we'd build it for you

What we build for accounting firms

  • The pain: Invoices keyed by hand into AP

    What we build: Invoice reader + vendor match + approval queue

    Published proof: −50% processing time, ≈0% errors (CA firm case)

  • The pain: K-1s and partnership packages swamp tax season

    What we build: Tax-document extraction into your tax software's import format

    Published proof: Faster delivery, less manual keying (RSM case)

  • The pain: Seniors stuck on statement analysis and tax research

    What we build: AI agent that drafts the vertical analysis and research memo for review

    Published proof: −75% execution time (peer-reviewed study)

Industry benchmarks

Where the profession is: interest is high, investment is low

Sentiment vs investment

  • Intrigued or excited by AI82%
  • Believe bookkeeping will be most disrupted59%
  • Believe firm value drops without AI54%
  • Actively investing in AI training25%
Source: Karbon, State of AI in Accounting 2024. Third-party published result.

Usage vs investment

  • Have used AI to help clients98%
  • Actively investing in AI training25%
Sources: 2024 Intuit QuickBooks Tech Survey (98%); Karbon 2024 (25%). Third-party published result.
  • 82%

    of accountants are intrigued or excited by AI

    Karbon - State of AI in Accounting 2024

  • 54%

    believe firm value drops without AI

    Karbon - State of AI in Accounting 2024

  • 59%

    believe bookkeeping will be most disrupted by AI

    Karbon - State of AI in Accounting 2024

  • 98%

    of accountants have used AI to help clients

    2024 Intuit QuickBooks Tech Survey

  • Only 25%

    are actively investing in AI training

    Karbon - State of AI in Accounting 2024

  • $6.68B → $37.6B

    AI in accounting market growth (2025→2030)

    Mordor Intelligence - 41% CAGR

Next step

Pick one workflow. We'll automate it in your current stack.

Bring your most manual workflow: AP intake, K-1 extraction, bank reconciliation or month-end analysis. We map it, build the automation on top of QuickBooks, Xero, your ERP or a portal without an API, and keep a human approval step. Typical first deployment: 6-12 weeks.

Works with what you run today

One automation layer connects to QuickBooks, Xero, your ERP, and portals with no API through browser automation.. ATI Lab automation layer (Approval steps agreed with you at scoping), which updates QuickBooks (API), Xero (API), Your ERP (API or browser automation), No-API portal (Browser automation).