Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Build, Buy, or Disappear: The AI Playbook for U...

Avatar for Fresh Books Tree Fresh Books Tree
August 31, 2026
1

Build, Buy, or Disappear: The AI Playbook for UK Accounting Practices

Explores how AI is transforming the UK accounting profession and the strategic choices firms must make to remain competitive.

Covers AI in UK accounting, accounting automation, Agentic AI, digital transformation, practice valuation, SaaS, and emerging competition from neobanks, AI-native firms, software platforms, and PE-backed consolidators.

A strategic resource for accounting professionals, firm owners, partners, CFOs, and business leaders navigating the future of accounting.

Avatar for Fresh Books Tree

Fresh Books Tree

August 31, 2026

Transcript

  1. Build, Buy, or Disappear The AI Playbook for UK Accounting

    Practices It is not a technology decision. It is the strategy that determines your survival and your value. › Two clear strategies for every practice › The Client Journey recurring revenue model › The PE-backed firm valuation playbook › The 5-Year Window — act now Richard Plasek MBA, FCMA, CGMA CFO • Board Director Henley Business School MBA Founding Member, Xeinadin Group • antler.london
  2. Build, Buy, or Disappear The AI Playbook for UK Accounting

    Practices A StrategicGuideforPracticeLeadersandPEInvestorsNavigatingtheAIEra AI will redefine every role, every client relationship, and every valuation in UK accounting. This is that moment. It is not a technology decision — it is the strategy that determines your survival and your value. Richard PlasekMBA, FCMA,CGMA CFO | BoardDirector | Henley Business School MBA Founding Member, Xeinadin Group
  3. About the Author Richard Plasek MBA, FCMA, CGMA is a

    senior finance executive with over 20 years of experience scaling professional services, energy, and advisory businesses across the UK and Europe. His career spans BigFour consulting at EY Capgemini and PwC, a successful public practice founded and grown over 23 years, and transformational board-level leadership of some of the UK's most ambitious professional services businesses. Richard was a founding member of Xeinadin Group — an overnight merger of more than 90 accounting firms across the UK and Ireland that has grown into a top-tier PE-backed UK accounting firm. His current roles include CFO and Board Member in international organisations where Richard advises on financial strategy, reporting, and business model optimisation with a genuinely international perspective. He holds an MBA from Henley Business School and a first-class BA (Hons) from the University of South Wales, as well as having contributed to the Financial Times and CIMA thought leadership on finance transformation.
  4. Preface There isamomentinthetransformationofeveryindustrywhentheoldandnewmodelscoexist—whenforwardthinkingpractitionerscanalreadyseeclearlywherethingsaregoing,whilethemajorityarestilldebatingwhether they needtopayattention.InUKaccounting,thatmomentisnow.Notapproaching.Now. AI will redefineeveryroleinthisprofession.Itwillredefineeveryclientrelationship.Anditwillmakethetraditional valuationmodels—theEBITDAmultiplesandGrossRecurringFeeformulasthathaveanchoredaccounting firm

    transactionsfordecades—insufficientontheirown.Thisisnotatechnologyevent.Itisastrategicevent. And forthosebuildingorinvestinginaccountingpractices,itisavaluationevent:thepracticesthatbuild proprietaryAgenticAIinthenextfiveyearswillnotsimplybebetterpractices—theywillbefundamentallymore valuable ones. The window to act is open now. It will not stay open. I havespentmorethan20yearsinaccountingandprofessionalservices—asaBig-4consultant,asapractice founder,asthearchitectofoneoftheUK'slargestovernightaccountingmergers,andasaboarddirectoradvising PE-backedbusinessesthroughcomplexgrowthandcapitalevents.Acrossallofthoseroles,Ihavewatched transformationssucceedandfail.Thesinglemostconsistentfindingisthis:thepracticesandbusinessesthat thrive are not those with the best technology. They are those with the clearest strategy, the most honest selfassessment, and the discipline to act before the window closes. Thefive-yearwindowreferredtothroughoutthisbookismystrategicassessment—basedonthecompetitive forcesmappedinthechaptersthatfollow—notanindustry-standarddefinition.Itreflectsthepaceatwhich platformsoftwarehouses,neobanks,PE-backedconsolidators,andAI-nativestart-upsaresimultaneously compressingthemarketfromeverydirection.TheconstrainttodayisnotAIcapability—itisphysical infrastructure.Butthatconstraintwillease.Thewindowcouldclosefasterthanfiveyears.Treatitasshorterthan you think, not longer. Thisbookdistinguishesbetweentwoclearstrategicpathsbecausetreatingallaccountingpracticesasifthey face the same choices leads to generic advice that serves no one well. A two-partner local practice and a PEbackedconsolidatorwith£250millioninrevenuedonotfacethesamestrategicquestion.Theyfaceopposite versionsofthesamestructuralshift—andtherightresponseforonewouldbethewrongresponsefortheother. StrategyAisthepathforindependentandsmallerpractices:buythebestavailabletools,connectthemtightly, automatetheback-office,andinvestfreedcapacityinthedeepadvisoryrelationshipsthatnotechnologycan replicate.StrategyBisthepathforPE-backedandlargeconsolidatedpractices:buildproprietaryAgenticAI trainedonyourowndata,developaSaaSplatform,andpositionfortheexitthatthetraditionalmodelcanno longer command. Thethreatlandscapehasexpandedbeyondwhatmostpractitionersaretracking.Thecompetitionisnolonger justfromotheraccountingfirms.Neobanksarebuildingaccountingintobanking.Platformsoftwarehousesare rebuildingthemselvesasAI-firstsystemsindirectpartnershipwiththemostpowerfulAIcompaniesintheworld. AI-nativestart-uppracticesareemergingwithnear-zerooverheadandthecapacitytoservehundredsofclients simultaneously.Understandingallfourcompetitiveforces—andwhereeachonecanandcannotberesisted— is thefoundationofanycrediblestrategicresponse. Crucially,AIdoesnotsellindividualservicelines.ItdoesnotbecomebetteratmarketingbookkeepingorVAT returnsasseparateproducts.WhatAgenticAIdoesisseetheclient'sjourney—itreadswhereeachclientisin theirgrowth,identifieswhattheyneedatthatmoment,andsurfacestheappropriateserviceautomatically.The ClientJourneyframeworkintroducedinthisbookisacommercialresponsetothatreality:amodelforbuildinga practicethatgrowswithitsclientsacrosseverystageoftheirdevelopment,creatingthekindofrecurringrevenue and switching costs that no competitor can easily displace. Thisbookisastrategicguide,notatechnicalmanual.Itdoesnottellyouwhichlinesofcodetowriteorwhich serverstoconfigure.Ittellsyouwhichdecisionstomake,inwhichorder,withwhichstrategicintent—andwhy thedecisionsyoumakeinthenexttwotothreeyearswilldetermineboththecompetitivepositionandtheexit valuation of your practice for the decade that follows. Richard Plasek MBA, FCMA, CGMA London, 2026
  5. Contents About the Author ................................................................................................................................................... 2 Preface .................................................................................................................................................................. 3 Contents

    ................................................................................................................................................................ 4 The AI Landscape: Two Strategies, One Profession ............................................................................................ 7 1.1 The Old Model vs the New Model ................................................................................................................ 7 1.2 Two Distinct Strategies — Clearly Distinguished ......................................................................................... 7 1.3 Three Categories of AI in Accounting .......................................................................................................... 8 1.4 The Two Strategic Paths Explained ............................................................................................................. 9 1.5 Two Types of AI: Generative and Agentic — and Why the Distinction Matters ........................................... 9 1.6 The Strategic Window for AI Advantage .................................................................................................... 10 1.7 Category 1 — Back-Office AI: From Referral to First Invoice .................................................................... 11 Diagnosing AI Readiness ............................................................................................................................. 11 The Integrated Client Journey ...................................................................................................................... 11 1.8 The People Management Layer: The Most Overlooked Inefficiency ......................................................... 12 AI Fluency as a Performance Metric ............................................................................................................ 12 1.9 The Partner's New Best Friend: From Secretary to AI Agent .................................................................... 13 1.10 The AI-Native Start-up: A New Competitive Force .................................................................................. 13 1.11 The Recruitment Imperative: Why PE-Backed Firms Must Not Freeze Hiring ........................................ 13 1.12 Category 1 — The Strategic Summary .................................................................................................... 14 Strategy A: Buy, Connect, and Automate .................................................................................................... 14 Strategy B: Build, Integrate, and Scale ........................................................................................................ 14 Category 2 — Work-Related AI: The Service Delivery Revolution ...................................................................... 15 2.1 The Full Service Range — AI Impact by Service Line ............................................................................... 15 2.2 Four Paradigm Shifts That Only Agentic AI and a Proprietary Model Can Deliver .................................... 16 Paradigm Shift 1 — Management Accounts: From Preparation to Prompted Advisory Conversation ......... 16 Paradigm Shift 2 — R&D Tax Relief: From Reactive Enquiry to Proactive Entitlement Identification ......... 17 Paradigm Shift 3 — Cross-Selling: From Forgotten Specialist Visits to AI-Driven Revenue Identification .. 17 Paradigm Shift 4 — Blockchain Accounting: The Fully Automated Service Line and the Coming Avalanche ..................................................................................................................................................................... 18 2.3 Why People Are the Constraint, Not the Technology ................................................................................ 18 The Four-Way Competition: Who Is Fighting Whom, and for What .................................................................... 19 3.1 Competitive Force 1 — Neobanks: The Banking Relationship as Accounting Relationship ..................... 19 3.2 Competitive Force 2 — Platform Software Houses: The Giants That Are Building ................................... 20 3.3 Competitive Force 3 — Strategy B Consolidators: The Firms Scaling Faster Than You .......................... 20 3.4 Competitive Force 4 — AI-Native Start-up Practices: The 1 Staff, 1,000 Client Model ............................. 21 3.5 The Competitive Map — At a Glance ........................................................................................................ 21 3.6 The Resolution: How Competition Becomes Partnership .......................................................................... 21 3.7 Strategy B's Missing Piece: The Micro-Business Brand ............................................................................ 22 The Client Journey: A Recurring Revenue Model for the AI Era ......................................................................... 23 4.1 The Client Journey Framework .................................................................................................................. 23 4.2 Stage 1 — Start-Up: The Most Contested Battleground ............................................................................ 23 Strategy A at Stage 1 ................................................................................................................................... 24
  6. Strategy B at Stage 1: The Micro-Business Brand ....................................................................................... 24

    4.3 Stage 2 — Growing SME: The Subscription Layer .................................................................................... 24 The Subscription Model at Stage 2 .............................................................................................................. 24 4.4 Stage 3 — Established Business: The Advisory Layer .............................................................................. 25 4.5 Stage 4 — Scale and Exit: The Strategic Tier ........................................................................................... 25 4.6 The Recurring Revenue Model: The Economics of the Full Journey ........................................................ 25 4.7 From Speed to Automation: Why Human Memory Cannot Be the System ............................................... 26 The Practical AI Stack for a Modern UK Practice ................................................................................................ 27 5.1 The Strategy A AI Stack: Buy, Connect, and Keep Current ...................................................................... 27 5.2 Connecting the Strategy A Stack ............................................................................................................... 28 5.3 The Strategy B AI Stack: Build, Not Buy .................................................................................................... 28 The Creativity Challenge: Reshaping Accounting Firm Culture .......................................................................... 29 6.1 The Wrong Conversation ........................................................................................................................... 29 6.2 The Stereotype and Its Truth ..................................................................................................................... 29 6.3 The Three Staff Groups: An Honest Assessment ...................................................................................... 29 Group 1: The Adapters ................................................................................................................................. 29 Group 2: The Resistant Staff ........................................................................................................................ 29 Group 3: The Hidden Talent ......................................................................................................................... 29 6.4 AI Fluency as a Cultural Metric .................................................................................................................. 30 6.5 Career Progression in the AI Era: From Time Served to Capability Demonstrated ................................... 30 6.6 Fixing the Reward Structure: The Most Important Structural Change ....................................................... 30 The PE Playbook: Building AI as Competitive Advantage and Exit Value in the 5-Year Window ....................... 32 7.1 The Valuation Argument: Two Live UK Data Points .................................................................................. 32 7.2 The Investment Thesis in the AI Era .......................................................................................................... 33 7.3 The Proprietary AI Imperative .................................................................................................................... 34 7.4 The 5-Year Window: A Deadline, Not a Guideline ..................................................................................... 34 7.5 What the Board Should Be Asking — At Every Quarterly Review ............................................................. 35 On Proprietary AI Development ................................................................................................................... 35 On the SaaS Platform .................................................................................................................................. 35 On Talent and Culture .................................................................................................................................. 35 7.6 The Board-Level Action Plan ..................................................................................................................... 35 7.7 The AI-Era Acquisition Due Diligence Lens ............................................................................................... 36 Governance, Regulation, and Implementation: The UK & European Framework ............................................... 38 8.1 The Five-Phase Implementation Framework ............................................................................................. 38 8.2 The Regulatory Landscape: UK GDPR, EU AI Act, and the Dual-Track Challenge .................................. 39 UK GDPR — The Non-Negotiable Foundation ............................................................................................ 39 8.3 Professional Standards and Liability .......................................................................................................... 40 8.4 AML, KYC, and Client Due Diligence ........................................................................................................ 40 8.5 Regulatory Checklist for AI Implementation ............................................................................................... 40 Appendix A — Competitor Threat Tracker .......................................................................................................... 42 Section 1 — Platform Software Houses .......................................................................................................... 42 Section 2 — Big 4 AI Initiatives ........................................................................................................................ 42 Section 3a — PE-Backed Consolidators ......................................................................................................... 43 Section 3b — Large Independent Networks .................................................................................................... 43 Section 4 — AI-Native Start-up Practices ........................................................................................................ 44
  7. Appendix B — Glossary of Key Terms ................................................................................................................ 45 Appendix

    C — Further Reading & UK Resources ............................................................................................... 46 Professional Bodies ......................................................................................................................................... 46 Regulators ....................................................................................................................................................... 46 Selected AI Development References ............................................................................................................. 46 Current AI Tool Recommendations ................................................................................................................. 46
  8. CHAPTER 1 — THE AI LANDSCAPE: TWO STRATEGIES, ONE PROFESSION

    The AI Landscape: Two Strategies, One Profession This book is built around a powerful central argument: there is no single AI challenge that all UK accounting practices must navigate in the same way. Instead, the profession is facing a structural bifurcation — a fork in the road leading to two fundamentally different strategic destinations. Understanding your practice's path and committing to it decisively is essential for everything that follows. 1.1 The Old Model vs the New Model The transition from the traditional to the AI-enabled accounting model is already underway. The comparison in Figure 1.1 is not a prediction — it is a description of changes that are measurably in progress in the UK market today. Figure 1.1 — Traditional vs AI-Enabled Accounting Practice Firms that have made this transition consistently demonstrate three measurableadvantages: higher profitability per partner; faster growth in client portfolios; and scalability — the ability to servesignificantly more clients without proportional headcount growth. These advantages compound over time, meaningthe gap between early movers and late followers widens every year. 1.2 Two Distinct Strategies — Clearly Distinguished The single most important structural point in this book is that there are two fundamentally different AI strategies for accounting practices. In this book we call them Strategy A and Strategy B. Each strategy requires different investments, capabilities, business models, and timelines. Any firm unable to distinguish between them and choose deliberately risks strategic paralysis. Figure 1.2 — The Two Strategy Fork
  9. Author's Perspective — Richard Plasek MBA, FCMA, CGMA Having built

    a practice from scratch, merged it into Xeinadinovernight alongside 90 other firms, and then guided PEbacked growth strategy, I have operated in both strategic contexts. The errors I have seen most often come from Strategy A practices trying to follow Strategy B playbooks — overreaching their capital and capability — or from Strategy B practices underestimating the urgency of the threat to their existing client base from early AI movers. 1.3 Three Categories of AI in Accounting Before examiningeach strategy in detail, it is important to understand how AI fits within accounting practice across three distinct categories. Understanding the categories first makes it easier to see how Strategy A and Strategy B each respond to them differently — and why the same tool can represent a complete solution for one strategy and merely a starting point for the other. Category 1 — Back-Office AI covers a firm's internal operational layer: marketing automation, client onboarding, invoice generation, resource scheduling, HR processes, deadline monitoring, and all non-client-facing activities. This is the lowest-risk category and the best place to start. The technology is mature, professional liability exposure is minimal, and efficiency gains are immediate and measurable. Category 2 — Work-Related AI covers the full range of accounting services delivered to clients: bookkeeping, tax computation, audit automation, management accounts, financial statement drafting, payroll, R&D tax relief, forensic accounting, corporate finance, and all other specialist services the practice offers. It is the largest, most commercially significant, and most professionally sensitive of the three categories. All AI outputs must be reviewed by a qualified practitioner before client delivery — human oversight is non-negotiable. Category 3 — Client Software Provision includes the AI-enabled tools and platforms that practices recommend, implement, and support for clients: cloud accounting platforms, MTD-compliant record-keeping, cash flow forecasting, expense management applications, payroll software, reporting and analytics tools, and so on across the full range of client-facing technology a modern practice deploys. The accountant who helps a client select and embed the right technology becomes genuinely difficult to displace. Figure 1.3 — Three Categories of AI in Accounting Practices Figure 1.3 maps the three categories through which AI operates in an accountingpractice. But the diagram alone does not tell the full story — how a practice implements each category dependsentirely on which strategic path itis on. Table 1.1 maps both strategies against all three categories, with the prime objective for each strategy clearly stated.
  10. Table 1.1 — Strategy A vs Strategy B: AI Implementation

    Across the Three Categories Strategy A (Independent & Small Practices) Category 1 — Back-Office AI BUY: Select best available tools (Ignition, Karbon, Breathe HR, Otter.ai). Connect via Zapier. Automate onboarding, invoicing, HR, and compliance reminders. Target: ~20% overhead reduction. Category 2 — Work-Related AI BUY: Xero, QuickBooks, IRIS Tax, Dext, Fathom. Use AI to reduce time per client. Deploy freed time into advisory. Mandatory human review of all AI outputs. Category 3 — Client Software Provision RECOMMEND & SUPPORT: Become the leading sector-specific expert in one or two platforms. Implement, train, and support clients. Audit independence protects this service line. PRIME OBJECTIVE Buy, Connect, and Automate — invest freed capacity in niche advisory expertise and personal relationships. When the ceiling is reached, link to a Strategy B SaaS platform to access capabilities beyond what Strategy A tools can deliver. Strategy B (PEBacked & Large Consolidated Practices) BUILD: Develop proprietary Agentic AI agents on top of platforms via APIs. Every backoffice process generates training data. Architecture must feed forward into the Category 2 data layer. PRIME OBJECTIVE Build, Integrate, and Scale — develop proprietary Agentic AI that cannot be purchased from any vendor. Offer the platform to Strategy A firms via SaaS to gain market share, expand training data, and build ARR that transforms the firm's exit valuation from a professional services multiple to a technology platform multiple. BUILD: Proprietary Agentic AI models trained on firm's own client data. Enables: prompted management accounts action, proactive R&D identification, AIdriven cross-selling. Not achievable with IRIS/CCH alone. BUILD ON TOP: Proprietary client portal connecting to Xero/QBO/Sage via live APIs. AI-generated insights under firm's brand. SaaS licensing of this platform to Strategy A firms creates new ARR revenue stream. 1.4 The Two Strategic Paths Explained The accounting profession in the AI era is not a single market facing a single challenge. It is two markets that are fast diverging, each with its own logic, threats, and strategic imperatives. That is a Strategy A or a Strategy B path. The choice is real. The paths do not converge. Strategy A is the path for independent and smaller practices. The competitive advantage here is not technology — it is depth. Deep sector expertise, genuine personal relationships, and community presence that a national platform cannot replicate at scale. Strategy A practices should adopt the best AI tools available, use them efficiently to free capacity for advisory relationships, and invest that capacity in becoming the undisputed expert in a specific niche. Strategy B is the path for PE-backed and large consolidated practices. The competitive advantage here lies not in relationships but in infrastructure: data assets, proprietary Agentic AI capability, and the ability to serve any client at any scale through technology that smaller practices cannot afford to build. The risk for Strategy B is moving too slowly — letting the strategic window close while still depending on the same platforms as every other firm in the market. 1.5 Two Types of AI: Generative and Agentic — and Why the Distinction Matters Before examining how AI operates within each category,itisessential to understand a distinction that most UK accounting firms are currently missing — and that determines whether their AI investment creates genuine competitive advantage or merely replicates what their competitors are already doing. Most UK accounting firms, when they talk about AI, are talking about Generative AI. Generative AI produces content — text, analysis, documents, summaries — in response to a human prompt. You ask, it answers. Microsoft Copilot drafting a client email, Claude summarising a set of accounts, ChatGPT answering a technical tax question — these are all Generative AI applications. They are genuinely useful. They save time. And they are reactive — the AI does nothing until a human initiates. Agentic AI is fundamentally different. An AI agent takes sequences of actions autonomously, monitors data continuously, triggers follow-up actions without being asked, makes decisions based on what it observes, and improves its performance over time. The human does not initiate — the AI does. It is proactive, self-directing, and continuous.
  11. Table 1.2 — Generative AI vs Agentic AI: The Distinction

    That Determines Competitive Advantage How it works Initiated by Agentic AI The human asks a question or makes a request The AI monitors, detects, and acts independently Drafting a tax letter, summarising accounts, answering a compliance query Accounting examples Current UK adoption Where value lives Strategy approach Generative AI Responds to human prompts — reactive A Widespread — most firms experimenting with Copilot, Claude, or ChatGPT Efficiency gains for individual tasks Buy and use Generative AI tools as daily work assistants Generative AI is table stakes — every competitor has it Strategy B imperative Acts autonomously without prompting — proactive Detecting a client's cash decline and prompting a partner call; proactively identifying R&D qualifying activity; flagging a cross-selling opportunity Rare — primarily the Big 4 and a small number of AI-native start-ups Competitive advantage at scale — the AI works 24/7 across every client simultaneously Limited — can access basic agentic features through platforms like Xero AI Build proprietary Agentic AI trained on own data — this is the moat that cannot be replicated The strategic implication is direct. When UK firms announce they are'implementing AI' and point to a Microsoft Copilot licence, they are describing Generative AI adoption. This is valuable and should be done. But it is not the source of durable competitive advantage. The competitive advantagethat Strategy B firms must build — the prompted management accounts system, the proactive R&D identification engine, the cross-selling intelligence platform — is all Agentic AI. It cannot be purchased from any vendor.Itmust be built. Where UK Accounting Currently Stands Almostall current UK accounting firmAI activity is Generative AI — using AI tools to draft, summarise, and answer. This is necessary but not sufficient. The Wolters Kluwer 2025 Future Ready Accountant Report identifies three levels of Agentic AI maturity in accounting firms: Taskers (automate low-value repetitive tasks), Automators (run entire processes end-to- end), and Collaborators (provide intelligent guidance during complex workflows). Most UK firms are at the Tasker level. Strategy B firms must reach the Automator and Collaborator levels within the strategic window. 1.6 The Strategic Window for AI Advantage The phrase 'five-year window' used throughout this book refers to the author's strategic assessment — based on the competitive forces mapped in the chapters that follow — that UK accounting practices have approximately five years from the current moment of accelerating AI adoption to establish a durable competitive position before the market consolidates around early movers. This is not an industry-standard definition or established consulting orthodoxy. It is a considered judgement. The strategic window for AI advantage is open now — but it is closing. The firms that act decisively during this period will shape the next generation of UK accounting. Those that wait — not for AI to prove itself, but for AI to become easily purchasable off the shelf, a temptation that applies particularly to Strategy A firms — will find that the competitive gap established by early movers is structural: one that cannot be closed by effort alone. How long does the window remain open? The honest answer is: nobody knows precisely — including the AI companies themselves. The constraint today is not AI capability — it is physical infrastructure. The demand for server capacity, GPU processing power, and energy to run AI systems at scale is outpacing the speed at which that infrastructure can be built. This means the window could close faster than a five-year planning horizon suggests. Treat the window as closing sooner than you think, not later. The Strategic Window Is Not a Guarantee Thewindow for building AI competitive advantage is open now. It is closing. The rate of closure is uncertain — AI development is accelerating faster than most industry observers predicted, and the physical infrastructure constraints that currently limit deployment speed will ease over time. Treat the window as shorter than you think, not longer.
  12. 1.7 Category 1 — Back-Office AI: From Referral to First

    Invoice Diagnosing AI Readiness Smallaccounting practices are not inefficient because their partners are bad at business. If they are inefficient, it is because their business model was built on personal relationships — and personal relationships made formal processes feel unnecessary. The commercial case for Category 1 AI is straightforward: practices that have implemented back-office AI automation report overhead cost savings of approximately 20%, alongside materially better cash management and a significantly faster and more professional client engagement experience — from first contact to signed engagement letter and first invoice. The most important shift in thinking about Category 1 is from a list of separate tools to a connected journey. The question is not 'which software should we buy for onboarding?' The question is: what happens from the moment apotential client first makes contact, all the way through to the moment their first invoice is paid — and how much ofthat journey can be handled automatically? How Most Practices Handle a New Enquiry Today — The Broken Process Apotential client contacts the firm. A marketing departmentor specific partner logs the enquiry on a list so they can chase later. The partner, when they have time, responds — sometimes promptly, sometimes days later. If the enquiry is converted, a new team is engaged to perform the onboarding. KYC documentation is chased manually. The engagement letter is drafted, printed, signed, and scanned. The first invoice is raised weeks later, after work has already begun. This process is not integrated. It is fragmented, human-dependent, and slow. At every handover point, the risk of delay — and the risk of losing the client to a faster competitor — increases. The AI-enabled model collapses this entire journey to a sequence that completes in under an hour, without partner involvement, before work begins. The Integrated Client Journey FirstContact: The potential client contacts the firm via website, email, WhatsApp, or QR code. An AI assistant responds immediately — within seconds — acknowledging the enquiry, asking qualifying questions, and setting expectations. No human involvement is required at this stage. Qualification and Routing: Simple engagements are routed to an automated proposal flow. Complex or highvalue enquiries are flagged for direct partner contact, with a summary of what the prospect has already shared prepared and waiting. Automated Proposal: The AI generates a fixed-fee proposal based on the qualifying information, delivered to the prospect's inbox within minutes. Engagement Letter and E-Signature: On proposal acceptance, the engagement letter is generated automatically and sent for electronic signature. The entire process from first contact to signed engagement letter can be completed in under an hour. Onboarding Sequence: On signature, a structured onboarding sequence is triggered automatically — welcome message, KYC (Know Your Customer) documentation and Anti-Money Laundering (AML) checks, bank feed connection, and a clear timeline for the first deliverable. First Invoice and Payment Set-up: The first invoice is issued automatically. A direct debit or GoCardless link is included. Cash collection begins before work starts. Ongoing Automation: Recurring invoices issued automatically. Payment reminders sent without partner involvement. Deadline alerts sent to clients automatically.
  13. Table 1.3 — The AI-Enabled Client Journey: Integrated Tools and

    Automation from First Contact to First Invoice Journey Stage First contact & AI response Recommended Tool Tidio, Intercom, or WhatsApp Business + AI assistant Integration Connected to enquiry form and CRM Typeform or built-in Ignition questionnaire Qualification & routing Ignition — automated fixed-fee proposals Feeds directly into proposal tool Proposal generation Ignition or DocuSign + practice template Connected to engagement letter and e-sign Engagement letter & esignature SmartSearch or Verify 365 Triggers onboarding on signature KYC & AML checks Karbon or Senta Practice management Xero Practice Manager Connected to practice management Xero / MTD set-up Ignition + GoCardless Central hub — connects all other tools Auto-triggered on onboarding completion Invoicing & payment collection Karbon automated workflows + email Recurring billing from day one Ongoing reminders & alerts Connect to Strategy B proprietary Agentic AI platform when available No manual intervention required Link to Strategy B SaaS Integrate as soon as commercially viable — to access capabilities beyond what Strategy A tools can deliver Why No Strategy B Stack Is Shown Here Strategy B practicesdo not implement Category 1 as a collection of off-the-shelf tools. For PE-backed and large consolidated firms, every back-office process is an architectural decision — each tool must expose an API that connects to the firm's proprietary Agentic AI data layer. The Strategy B Category 1 implementation is addressed in Chapter 7, alongside the full proprietary AI development programme. 1.8 The People Management Layer: The Most Overlooked Inefficiency Small accounting practices haveno HR department. Inmost one- to four-partner firms, the senior partner is simultaneously the managing director, the finance director, the HR manager, and the office administrator. Staff primarily leave not because of pay, but because they feel undervalued, unrecognised, and unclear about their future. Key performance metrics for every fee earner should include: advisory hours billed against target — not total chargeable hours, which should be declining as automation takes over data processing; AI efficiency gains delivered; client satisfaction scores; new advisory relationships developed; and AI fluency score assessed quarterly. AI Fluency as a Performance Metric AI fluency should be assessed across four dimensions quarterly: tool adoption; prompt quality; output judgement; and workflow integration. For Strategy A practices, AI fluency is a Generative AI usage metric. For Strategy B practices, it extends into the ability to teach the Agentic AI — contributing to the firm's proprietary model. At its most advanced, this is a product development metric, not a training metric. Table 1.4 — The AI-Enabled People ManagementStack: Tools, Functions, and What They Replace Function Recommended Tool Otter.ai or Fireflies.ai What It Replaces Meeting recording & transcription Karbon tasks or Notion Handwritten minutes that nobody reads Action tracking & follow-up Karbon reporting + Google Sheets Actions written in notebooks that disappear Performance dashboards Breathe HR or Factorial Partner's informal impression HR records & leave management ICAEW/ACCA portals +Breathe HR Spreadsheets and paper forms CPD & development tracking Glassdoor Insights or ICAEW salary data Annual conversation at review time Salary benchmarking Officevibe or Culture Amp starter Partner's gut feeling Staff surveys & pulse checks Zapier Nothing — or annual anonymous feedback Integration layer Manual updates across multiple systems
  14. A Note for Strategy B Practices Thepeople managementtools in this

    section apply to Strategy A practices. For Strategy B practices, people management metrics — including AI fluency scores, performance dashboards, and CPD tracking — should be integrated into the firm's proprietary Agentic AI platform, feeding the same data layer as the client-facing systems. 1.9 The Partner's New Best Friend: From Secretary to AI Agent Everyfunction a practice secretary has traditionally performed can now be accomplished better, faster, more consistently, and at a fraction of the cost by an AI agent. Diary management. Correspondence drafting. Client chasing. Document filing. Meeting scheduling. Reminder sending. The AI does not take holidays. It does not miss a deadline. This is not a comfortable observation — but avoiding the decision does not make the economics less real. The AI does not make the partner's client calls. The partner does. But the AI ensures the call happens at exactly the right moment, with exactly the right information. When a client's cash position falls 34% in six weeks, the AI surfaces a prompt: 'Client X's cash position has deteriorated significantly. Corporation tax due in eight weeks. Last contact 43 days ago. Recommend a call this week.' The partner makes the call. The client feels valued. The relationship deepens. The Contrast With Strategy B Strategy B practices face a different version of this challenge. While Strategy A firms must make difficult decisions about individual administrative roles, Strategy B firms must simultaneously make those decisions AND avoid the opposite error: freezing recruitment hoping that headcount will naturally reduce through attrition. At a PE-backed firm, the secretary and manual administrator roles are displaced — those decisions must be made. But the AI-literate fee earners, the model trainers, and the advisory professionals the firm needs to build its proprietary Agentic AI platform must be recruited. The two decisions are not in tension. They are both necessary. 1.10 The AI-Native Start-up: A New Competitive Force Theaccounting profession faces four distinctcompetitive forces in the AI era — neobanks, platform software houses, PE-backed consolidators, and a fourth force that receives the least attention: the AI-native start-up practice. All four are examined in full in Chapter 3, which maps the complete competitive landscape. The AInative start-up is introduced here because it directly shapes how you should think about Category 1 urgency and client acquisition speed. The Number That Should Stop Every Partner in the Profession One staff member. A thousand clients. This is not a thought experiment. Docyt, the AI-powered accounting automation platform, has explicitly launched a mission to enable every accountant to become a 'Million Dollar Accountant' — serving 300 small businesses simultaneously at an average fee, with 60%+ gross margins, through Agentic AI automation. Digits has developed an Autonomous General Ledger that embeds AI agents to run entire accounting workflows from start to finish, pausing only when human judgement is necessary. The economics of the one-person AI-native practice are real, achievable today, and will become increasingly mainstream within the strategic window. The AI-native start-up has no legacy systems, no staff to retrain, no culture to shift, no billing model to reform. It isa competitive threat to Strategy A practices for clients — and to Strategy B practices for talent. The best young qualified accountants now have a genuine choice: join a firm, or build an AI-native practice with near-zero overhead. 1.11 The Recruitment Imperative: Why PE-Backed Firms Must Not Freeze Hiring The junior — whether a recently qualified accountant, an experienced bookkeeper, ora member of a pool of junior resources — performs the data processing layer of accounting. It is not a question of age or seniority in the traditional sense; it is a question of the work being performed. AI replaces specific tasks in that layer. It does not replace the judgement, relationships, and commercial thinking that make an accounting practice genuinely valuable. When a Strategy B accounting firm begins implementing AI, a conversation happens in almost every boardroom: AI will automate a significant proportion of junior and mid-level work; therefore we should slow or stop recruitment until the AI implementation is complete. This logic is seductive. It appears financially disciplined. And it is, in almost every respect, wrong.
  15. The error is a category mistake: confusing the automation of

    tasks with the elimination of roles. The correct approach is not to freeze recruitment and wait for natural attrition to reduce headcount. It is to change the composition of hiring — reducing recruitment for data processing and manual compliance roles while actively increasing recruitment for AI-fluent fee earners, model trainers, advisory professionals, and commercial developers. The accelerated partnership track described in Chapter 6 should be communicated explicitly to recruits at the point of hire. For AI-fluent candidates choosing between firms, this is a competitive differentiator that most firms are not yet articulating. The Closing Point AImultiplies people. It does not replace them. A firm with fifty AI-enabled fee earners can outperform one with a hundred traditional fee earners. But it is still a firm with fifty people — people who need to be recruited, developed, retained, and succeeded. The investment in people does not stop because AI has arrived. It becomes, if anything, more important. The firms that win in the AI era will not be those with the best technology. They will be those with the best people using the best technology. 1.12 Category 1 — The Strategic Summary Strategy A: Buy, Connect, and Automate For independent and smaller practices, Category 1 AI is an exercise in efficiency and professionalism. Use the best technology available for the Strategy A opportunity set, connect it tightly, and automate the back-office layer so that partners and fee earners spend the maximum proportion of their time on client-facing work. Strategy A firms are buyers, not builders. The ceiling of Strategy A in Category 1 is also clear: these tools are available to any practice. Deploying them creates efficiency but not competitive differentiation. When that ceiling is reached — when the practice can no longer grow through buying and connecting tools alone, and when competitive pressure from platforms and AInative start-ups intensifies — the natural next step is to connect to a Strategy B firm's SaaS platform. This is not a defeat. It is a graduation. Strategy B: Build, Integrate, and Scale For PE-backed and large consolidated practices, Category 1 is not an efficiency exercise. It is the foundation layer of a proprietary Agentic AI platform — and every decision made in Category 1 must be made with that larger objective in mind. Every tool deployed has a second purpose: it is a data source, a training input, and a building block for the proprietary Agentic AI system that will power competitive advantage in Categories 2 and 3. Buy and Connect, or Build and Integrate Theback-office is wherethe strategic choice between Strategy A and Strategy B becomes real. Strategy A practices should buy the best available tools, connect them tightly via Zapier or Make, and automate everything that does not require human judgement. Strategy B practices should build back-office systems architecturally connected to their broader Agentic AI development programme — feeding data forward into Category 2 and eventually into the proprietary clientfacing platform that defines their competitive position in the AI era.
  16. CHAPTER 2 — CATEGORY 2: WORK-RELATED AI — THE SERVICE

    DELIVERY REVOLUTION Category 2 — Work-Related AI: The Service Delivery Revolution In Chapter 1, we established the three categories of AI in accounting and introduced the distinction between Generative AI — which responds to prompts — and Agentic AI, which acts autonomously, monitors continuously, and improves over time. Category 2 is where the distinction between the two matters most, and where the difference between Strategy A and Strategy B becomes commercially decisive. Category 2 covers Work-Related AI: the delivery of accounting services to clients — bookkeeping, tax computation, audit automation, management accounts, financial statement drafting, payroll processing, and the advisory services above them. It is the largest, most commercially significant, and most professionally sensitive of the three categories. It is also where most of the market conversation about AI is focused. But before examining what is possible in Category 2, this chapter needs to state something plainly that the industry is reluctant to say out loud. The Warning That Opens This Chapter Buyingmajor compliance platforms — such as IRIS, CCH, or Sage, to name just a few examples — however AI-enhanced the marketing, is not an AI strategy for a PE-backed accounting firm. When a firm with £250 million in revenue approves a budget line for a platform upgrade and describes this as its AI readiness programme, it has not created competitive advantage. It has bought a Generative AI tool that its competitors have access to by lunchtime. Worse, in doing so, it may be inadvertently giving away its most valuable strategic asset — its client data — as training material for the vendor's own AI models. Every client transaction processed through a major compliance platform potentially enriches that platform's AI capabilities, not the firm's. PE-backed firms have the resources to overcome this — by building proprietary Agentic AI trained on their own data, kept within their own systems, generating competitive advantage rather than surrendering it. Strategy A firms have little choice but to use these platforms. But they should be aware of what they are giving, and to whom. The proprietary Agentic AI model trained on the firm's own data is not an optional enhancement. For Strategy B firms, it is the strategy. Everything else — every platform licence, every software subscription — is infrastructure. And infrastructure is not a competitive moat. 2.1 The Full Service Range — AI Impact by Service Line The following table maps AI's impact across the full range of services a comprehensive accounting practice delivers. A 'Strategy' column indicates whether each service primarily involves Strategy A practices, Strategy B practices, or both. Table 2.1 — AI Impact Across the Full Range of Accounting Service Lines Service Line Bookkeeping & bank reconciliation VAT returns & MTD submissions Payroll & RTI submissions Management accounts AI Automation Potential Near 100% Human Judgement Required Minimal Strategy Notes A/B MTD drives complete digitisation. Agentic AI categorises, reconciles, and flags exceptions continuously. Near 100% Minimal A/B Rule-based, legally mandated digital links. Near-full automation achievable today. Near 100% Minimal A/B Near 100% Low A/B Self-assessment tax returns Very high Corporation tax compliance Very high Medium Medium Very high Medium Capital allowances High High Audit & assurance High R&D tax relief A/B Medium A/B B A / B A/B Highly rule-based. Agentic AI handles calculation, submission, and exception flagging. Auto-calculated from live bookkeeping data. The constraint is not preparation — it is client action. See Paradigm Shift 1 below. Structured data input. AI drafts; practitioner reviews and approves. Calculated from accounts data. AI drafts computation; practitioner reviews. Agentic AI identifies qualifying activity proactively from client data. Practitioner validates. See Paradigm Shift 2. AI identifies assets and classifies expenditure. Practitioner applies judgement on marginal cases. Agentic AI analyses full transaction population. Practitioner exercises professional scepticism and judgement.
  17. Service Line AI Automation Potential Forensic accounting High Medium VAT

    advisory & indirect tax Medium Low Tax planning & advisory Medium Low Corporate finance & M&A Mediumhigh Grant funding advisory Human Judgement Required High High Strategy Notes B Agentic AI detects anomalies faster and at scale. Practitioner interprets, investigates, and presents findings. A/B A/B Very high Near 100% B Very high A/B High A/B Succession & IHT planning Very high ESG reporting Medium Blockchain accounting Minimal AI researches and drafts; practitioner advises on complex positions. AI models scenarios; practitioner leads the conversation and the strategy. AI supports analysis and documentation; practitioner leads relationships and negotiations. B Agentic AI identifies eligible programmes and drafts applications; practitioner refines and submits. B Deeply personal, relationship-based, judgement-intensive. AI gathers andanalyses data; practitioner ensures accuracy and advises on strategy. Agentic AI managesthe data layer entirely. Human role: tax treatment, regulatorycompliance, valuation advisory. See Paradigm Shift4. The strategic insight from this table is not that AI automates everything—itisthat Agentic AI automates the highvolume, structured, rule-based layer of accounting almost entirely, andsignificantly augments the judgementbased layer. The services that survive as genuinely human-led are exactlythe services that clients value most, pay most for, and find most difficult to obtain elsewhere. AI is not erodingthevalue of accounting. It is clarifying it: stripping away the transactional layer to reveal the advisory layer underneath. But here is the point that this table, by its very structure, risks obscuring:AIdoes not sell individual service lines. A practice that deploys Agentic AI does not become better at marketingbookkeeping, or VAT returns, or R&D tax relief as separate products. What Agentic AI does is see the client'sjourney — it reads where each client is in their growth stage, identifies what they need at that moment,andsurfaces the appropriate service automatically. The accountant is not selling a service line. The AIisrecognising an opportunity within a relationship. This is the shift that makes the Client Journey framework—firstreferenced later in this chapter and addressed in full in Chapter 4 — the commercial heart of the book. 2.2 Four Paradigm Shifts That Only Agentic AI and a Proprietary Model Can Deliver The following paradigm shifts are achievable only with a proprietary Agentic AI model trained on the firm's own client data. They require AI that acts without being prompted, monitors continuously, and improves over time. They are not achievable with Generative AI tools purchased from vendors — however sophisticated those tools may be. This section is therefore addressed primarily to Strategy B firms. Strategy A practices should read it to understand what is becoming possible, and to consider how access to a Strategy B firm's platform via SaaS might bring some of these capabilities within reach. Paradigm Shift 1 — Management Accounts: From Preparation to Prompted Advisory Conversation The management accounts problem in UK accounting practice is not a preparation problem. The preparation layer is effectively solved — cloud accounting platforms, AI-assisted categorisation, and automated reporting mean that management accounts can be prepared from live bookkeeping data in minutes, not days. The problem is the last mile. The client who receives the management accounts pack every month, glances at the summary page, and carries on doing exactly what they were doing before. The accounts have been prepared. The information has been delivered. And nothing has changed. There is also a timing problem that even the best traditional management accounts model fails to address: management accounts are periodic and post-event. Changes at the beginning of a month are not highlighted until approximately four weeks later. By then, what appeared to be a structural issue may have turned out to be seasonal variation — or what appeared to be seasonal may have been structural. Agentic AI resolves this by monitoring the client's financial data in real time, continuously, detecting changes the moment they occur. A proprietary Agentic AI model changes the management accounts model fundamentally. It does not simply prepare the accounts and send them. It prepares them from live data, reads them immediately, compares them against the client's historical trends and sector benchmarks, identifies the most significant issues, and triggers a prompted advisory conversation. The client might receive a message such as the following — the precise wording and tone should reflect each firm's own practice and the preferences of the individual client:
  18. Example — Prompted Advisory Conversation 'Your accountant has reviewed this

    month's management accounts and would like to discuss three areas with you. The most significant is a 6.4% decline in gross margin over the past quarter, which may be driven by pricing pressure, rising supplier costs, or a change in your sales mix. Your management accounts include a brief summary of the key figures and a drill-down into the areas of concern. We would welcome a short conversation to talk through the options — please use the link below to confirm, reschedule, or let us know if you would prefer to handle this by email instead.' Each firm's approach to client communication will differ. Some clients will welcome a proactive scheduled call. Others will prefer to respond in their own time. The important principle is that the AI has identified the issue, provided the data, and created the opportunity for a conversation — rather than leaving the client to notice the problem themselves, or not notice it at all. Where the handover from AI to human accountant should occur is a decision for each firm, informed by client preferences and the nature of the issue identified. The Commercial Argument — And the Real Advance Over Traditional Management Accounts The samemanagement accounts that previously generated a fee andafiling have now generated a prompted advisory conversation, a strategic recommendation, and — in many cases — additional advisory work that the client did not know they needed. The compliance layer has funded the advisory layer. Automatically. At scale. But the deeper advance is this: management accounts have always been a post-event document. By the time the client receives them, the events they describe are already four weeks old. Agentic AI changes the timeline entirely. It monitors continuously, detects changes as they occur, and can trigger a conversation within days of a significant movement — not weeks. A cash position that begins deteriorating in week one of a month need not wait until the month-end accounts pack to be identified. The AI sees it, flags it, and prompts the conversation while there is still time to act. Paradigm 2— Relief: Reactivepractice Enquiryis to Proactive Entitlement Identification The currentShift model forR&D R&DTax tax relief in From UKaccounting broken in a specific and commercially significant way. The accounting firm asks clients whether they do any R&D — typically at the annual review meeting, typically as one item on a long agenda, typically without sufficient context for the client to give a meaningful answer. The client says no, because they think R&D means scientists in white coats. Billions of pounds of legitimate R&D tax relief goes unclaimed every year. HMRC's 2023 reforms — which merged the RDEC and SME R&D schemes, introduced mandatory prenotification, and significantly tightened documentation requirements — made legitimate R&D claims harder to prepare and even more dependent on proactive identification and careful evidence-gathering. A proprietary Agentic AI model flips this entirely. Instead of asking the client whether they do R&D, the AI analyses what the client actually does — reading their bookkeeping entries, payroll data, project codes, supplier invoices, bank transactions, and any other structured data available through the Category 2 service delivery layer. It identifies patterns consistent with qualifying R&D activity and surfaces a prompt to the partner with the estimated value of the potential claim. The client has not asked about R&D. The partner has not had to remember to ask. The AI has done the identification, estimated the value, and created the opportunity. For a Strategy B firm with a large and diverse client portfolio, this is not just a compliance improvement — it is a new revenue stream, a new advisory relationship, and a tangible demonstration of value that no Generative AI tool and no off-the-shelf platform can replicate. Paradigm Shift 3 — Cross-Selling: From Forgotten Specialist Visits to AI-Driven Revenue Identification Every accounting firm above a certain size has experienced a version of the following scenario. The R&D team visits regional offices. The area director says 'talk to my account managers.' The account managers attend the presentation. Three months later, nothing has happened. The opportunity has evaporated — not because it did not exist, but because the system for acting on it depended entirely on individual memory. Research from Accounting Today confirms that cross-selling to existing clients accounts for up to 30% of revenue for professional services firms that execute it well. McKinsey's research across multiple industries found that cross-selling accounts for approximately 20-21% of the value companies derive from revenue synergies — yet fewer than 20% of organisations achieve their cross-selling goals. For a Strategy B firm with £250 million in revenue, the unrealised cross-selling value within the existing client portfolio — that is, from existing clients, not prospects — is between £50 million and £62 million per year. A proprietary Agentic AI model changes this from a human initiative problem to a systems problem. The AI reads every existing client's financial data across every service line currently delivered, cross-references it against the firm's full service range, and creates specific actionable prompts: the manufacturing client receiving bookkeeping and corporation tax compliance who has not been assessed for R&D eligibility; the property developer not offered capital allowances analysis; the owner approaching retirement who has not been introduced to the succession planning team.
  19. The Cross-Selling Argument — Stated Plainly The AIhas alreadydone the

    matching,created the opportunity, and assigned the accountability. In a firm of £250 million revenue, a 20% improvement in cross-selling conversion — driven by systematic Agentic AI identification rather than individual human memory — is worth £50 million in additional annual revenue from existing clients. That is achievable only by a firm that has built the Agentic AI infrastructure to see its entire client portfolio as a connected data asset. Paradigm Shift 4 — Blockchain Accounting: The Fully Automated Service Line and the Coming Avalanche Blockchain accounting represents something genuinely new: a service line whereAgenticAI can manage the data layer almost entirely, with minimal human involvement in the processing layer. A blockchain creates an immutable, continuously updated, tamper-proof record of every transaction the moment it occurs. The accounting entry is not made after the transaction — it is the transaction. For practices with clients operating in crypto, DeFi, Web3, tokenised assets, or any blockchain-native environment, Agentic AI reads, interprets, reconciles, and reports on blockchain transaction data in real time, continuously, without human intervention. The human accounting role in blockchain accounting shifts entirely: to tax treatment advisory, regulatory compliance, valuation advisory, and strategic counsel on the tax-efficient structuring of blockchain-based business models. But blockchain accounting is not merely a niche service for crypto clients. It represents a structural shift in how accounting itself works — and when it arrives at scale, it will not arrive gradually. The likely trajectory is one of accelerating change: slow initial adoption followed by rapid, widespread disruption as the infrastructure matures and as businesses and regulators converge on common standards. The Blockchain Reckoning for Small Practices When blockchain accounting reaches mainstream adoption, the compliance processing layer that currently sustains thousands of small UK accounting practices will largely cease to exist as a billable service. Transactions will record themselves. VAT positions will always be current. Year-end accounts will be a formatted version of what the system already knows. The practices that survive will be those that have moved decisively to the advisory layer — or those that have linked to a PE-backed group that is blockchain-ready. This is not a distant scenario to plan for eventually. It is a structural shift to begin preparing for now. Forthcoming Publication Thefull strategic and technical implications of blockchain for accounting practice — including the Agentic AI architecture required to manage blockchain accounting at scale, the regulatory framework as it develops in the UK and EU, and the commercial opportunity for practices that develop genuine expertise in this area — are the subject of a forthcoming publication from Antler Associates. 2.3 Why People Are the Constraint, Not the Technology Everything described in this chapter is technically achievable today. The technology exists. The data exists within the firm's own systems. The commercial case is clear and, in many instances, quantifiable. What is not readily available — and what cannot be purchased from any vendor — is the human capability to build, train, operate, and continuously improve these systems. The people required to build a proprietary Agentic AI model in an accounting context are not generic software engineers. They are a specific and rare combination: technically fluent in AI systems, deeply knowledgeable about UK accounting practice and tax law, commercially minded enough to understand which problems are worth solving, and experienced enough to know where the professional judgement boundaries lie. The five-year strategic window, in Category 2, is above all a talent window. The technology can be built in eighteen months bythe right team. Assembling that team takes longer. The firms that commit now will have built proprietary modelswith two years of training data by the time the window closes. The firms that begin later will be starting from scratch, competing for talent that the early movers have already secured. The Closing Point Notawindow to buy better software. A window to build something that cannot be bought. Category 3 — Client Software Provision — follows in Chapter 3 as part of the competitive landscape map. The full Category 3 strategic framework, including the API imperative and the SaaS resolution, is addressed in the Four-Way Competition chapter.
  20. CHAPTER 3 — THE FOUR-WAY COMPETITION: WHO IS FIGHTING WHOM,

    AND FOR WHAT The Four-Way Competition: Who Is Fighting Whom, and for What The accounting profession in the UK is not facing a single competitive threat. It is facing four — simultaneously, from different directions, at different speeds, and targeting different parts of the market. This chapter maps those four competitive forces and, crucially, identifies which firms they threaten most, what they are taking, and what remains defensible. Understanding this map is the foundation of a coherent competitive strategy — whether your practice is on Strategy A or Strategy B. Figure 3.1 — Platform Threat Map The four forces are: neobanks; platform software houses; StrategyBconsolidators; and AI-native start-up practices. Together they represent a compression of themarket fromevery direction simultaneously. But they are not equally threatening to every practice, and the strategic responseavailable to each type of practice differs fundamentally. That distinction — the Strategy A response and the Strategy B response — runs through every section of this chapter. 3.1 Competitive Force 1 — Neobanks: The Banking Relationship as Accounting Relationship The first and most immediate competitive threat to Strategy A practices is not another accounting firm. It is the neobank. ANNA Money, Revolut Business, Monzo Business, and FreeAgent — now owned by NatWest — have identified a structural vulnerability in the accounting profession: the firm's relationship with its smallest clients is built almost entirely on compliance services that can be automated. These are precisely the services that neobanks are now offering as a bundled extension of the business banking relationship. The neobank's structural advantage is not primarily technological. It is relational. The neobank already has the client's banking data. The accounting service is an extension of an existing relationship, built on data the neobank already holds, delivered through an app the client uses every day. Table 3.1 — Neobank Competitors: Current AI Accounting Capabilities Neobank ANNA Money Revolut Business Current AI Accounting Capabilities Receipt/invoice upload, automated bookkeeping, MTD VAT returns within banking app, payroll expanding Who They Target Sole traders, micro-businesses, startups Expense management, invoicing, bookkeeping integration and tax reporting expanding SMEs, freelancers, growing businesses Integrated accounting through Xero/QBO partnerships, expense categorisation Small businesses, sole traders Monzo Business FreeAgent (NatWest) Tide Full MTD-compatible bookkeeping, VAT returns, payroll, self-assessment — free with NatWest accounts Receipt capture, bookkeeping via Xero integration, VAT returns, invoice management, AI expense categorisation Small businesses banking with NatWest/RBS/Ulster Bank Small businesses, sole traders, contractors
  21. What neobanks cannot take — and this is Strategy A's

    defensible position — is the advisory relationship. The complex tax question. The R&D claim. The business valuation. The succession plan. The conversation about whether to take on a new partner, raise debt, or sell the business. These require human judgement, professional expertise, and a depth of client understanding that no banking app can replicate. Strategy B firms do not lose clients to neobanks in the same way. But a failure to invest in a separately branded micro-business and start-up service tier — priced and positioned to compete with neobank offerings — means losing the pipeline: the start-up clients of today who would have become established business clients of tomorrow. 3.2 Competitive Force 2 — Platform Software Houses: The Giants That Are Building The second competitive force is larger, better resourced, and more far-reaching than the neobank threat. It is the platform software houses — and unlike the neobanks, which are targeting the start-up and micro-business segment, the platforms are targeting the entire compliance layer of accounting across every client segment simultaneously. Xero, QuickBooks, Sage, Oracle, and SAP are not adding Generative AI features to existing products. They are rebuilding their platforms as AI-first systems, in direct partnership with Silicon Valley firms — Anthropic, OpenAI, Microsoft, Google — that are developing the most powerful AI models in the world. They are investing billions. They are training their models on the transaction data of millions of businesses globally. The Observation This Book Needs to State Plainly These platformsare not buying AI.They are building it — exactly as this book argues Strategy B accounting firms should be doing. The irony is sharp: the platforms that Strategy B accounting firms are currently buying instead of building their own AI are themselves building exactly the kind of proprietary Agentic AI that this book argues Strategy B firms should be building. And they are building it with training data from millions of businesses — far more than any individual accounting firm can assemble. Furthermore, every firm that buys a licence to these platforms and processes client data through them is potentially contributing to that training dataset. The accounting firm that buys an IRIS or CCH licence is not just a customer. It may, depending on the vendor's terms, be a data supplier. In the United States, AI systems are already automatically preparing partnership tax returns — documents that in the UK context require detailed professional knowledge and significant technical expertise. Whether this specific capability arrives in the UK market in two or five years is not the critical question. The question is whether your practice is building the advisory capability that will matter when it does. This is a clear manifestation of a broader trend: the traditional services offered by small accounting firms — bookkeeping, VAT, payroll, basic tax compliance — are now being offered for free or near-free by software platforms, and increasingly automated without any human intervention. For Strategy A practices, this is the most important structural shift to understand: the compliance services that sustained the traditional model are being systematically removed from the billable zone. 3.3 Competitive Force 3 — Strategy B Consolidators: The Firms Scaling Faster Than You The third competitive force is the one that Strategy A practices find most uncomfortable to acknowledge, because it comes from within the profession. The PE-backed consolidators — Xeinadin, Azets, Sumer, Evelyn Partners, Grant Thornton (Cinven-backed), Cooper Parry, and the growing cohort of similar firms — are not simply larger accounting practices. They are professionally managed, PE-backed businesses that are systematically acquiring the client relationships and market presence that Strategy A practices have built over decades. Strategy B firms are targeting Strategy A's best clients — the growing SMEs, the established owner-managed businesses, the mid-market companies approaching an exit event. They offer national scale, specialist teams, technology investment, and brand credibility that carries weight with the professional intermediaries — lawyers, banks, corporate finance advisers — who refer the most valuable clients. What Strategy B firms cannot take from Strategy A practices — and this is the genuine, durable competitive advantage that Strategy A must invest in defending — is the depth of personal relationship, the genuine local knowledge, and the trust built over years and decades in a specific community and sector. The client who has worked with the same partner for fifteen years, whose business the partner genuinely understands from the inside, is not easily moved by a competitor's marketing campaign. But they are moveable — if the Strategy A practice fails to provide the specialist services the client now needs, or if the responsiveness and technology quality gap becomes too wide to ignore. The full competitor intelligence for all major PE-backed consolidators — including current revenue, PE backer, and AI/technology strategy — is provided in Appendix A, Section 3a.
  22. 3.4 Competitive Force 4 — AI-Native Start-up Practices: The 1

    Staff, 1,000 Client Model The fourth competitive force receives the least attention because it does not yet have a name, a trade association, or a representative body. It is the AI-native start-up practice — a single qualified accountant, working alone, serving a client base that would have required a team of eight to ten people under the traditional model. The economics of this model — validated by US platforms such as Docyt and Digits — are examined in full in Chapter 1, Section 1.10. It is included here because it is the fourth of the four competitive forces that together define the market pressure on UK accounting practices. The AI-native start-up practice has no legacy systems, no staff to retrain, no culture to shift, no billing model to reform. It is not just a competitive threat to Strategy A practices for clients — it is a competitive threat to Strategy Bpractices for talent. The best young qualified accountants now have a genuine choice that did not exist five years ago: join a firm, or build an AI-native practice with near-zero overhead. 3.5 The Competitive Map — At a Glance Thefollowing table summarises the competitive landscape across the four stages of the Client Journey — the framework that maps a client's progression from start-up through to exit, with AI-enabled services, competitive threats, and strategic responses at each stage. As noted in Chapter 2, AI does not sell individual service lines across these stages. It sees the client's position in their journey and surfaces the appropriate service at the appropriate moment. The Client Journey framework is addressed in full in Chapter 4. Read the table below by client stage rather than by service. Each row follows the same business as it grows, showing where the sharpest threat sits at that point, what is most at risk of being taken, and the position each strategy can defend. The pattern that emerges is the argument of this chapter in miniature: the threat changes at every stage, but so does the response available to it. Table 3.2 — The Competitive Map: Four Forces Across the Client Journey Client Stage Stage Primary 1: Start-Up (£0- Threat 250k turnover) Neobanks + AI-native Stage 2: Growing start-ups SME (£250k-£2m turnover) Platform software + Strategy B Secondary Threat Platform software What Is Being Taken Bookkeeping, VAT, basic payroll — the compliance entry layer Defensible Position Speed of response; niche sector expertise; Strategy B micro-brand Strategy A or B A: compete on speed and expertise. B: launch micro-brand AI-native start-ups scaling up Compliance processing, management accounts, cash flow reporting Deep advisory relationship; Agentic AI prompted conversations; specialist services A: advisory depth. B: proprietary AI advisory engine Stage 3: Established (£2m£20m turnover) Strategy B consolidat ors Platform enterprise software Stage 4: Scale/Exit (£20m+) Strategy B consolidat ors + Big 4 Platform enterprise software Mid-market advisory, specialist services, M&A support Strategic advisory, capital markets support, board-level CFO services Personal relationship depth; audit independence; local knowledge; trusted adviser status Track record; personal trust; relationship continuity through transactions A: relationship and audit. B: virtual CFO platform B: premium advisory. A: refer to B partner 3.6 The Resolution: How Competition Becomes Partnership The competitive map described above is genuinely threatening. But it is not without resolution — and the resolution, when it is found, is more commercially powerful than any defensive strategy. The resolution is this: Strategy B firms should not be competing with Strategy A firms for clients. They should be enabling them. A Strategy B firm that successfully builds a proprietary Agentic AI platform — one that monitors clients continuously, identifies opportunities, prompts conversations, and learns from outcomes — has created something Strategy that every Strategy A firm in the country needs and cannot build for itself. This does not mean B firms should ignore the start-up and micro-business segment entirely. As discussed in Section 3.7, this book recommends that Strategy B firms establish a separately branded micro-business tier — precisely for those clients who genuinely want to be with a Strategy B practice from the start, and who will grow into the premium service tier over time. The micro-brand is not competition with Strategy A. It is a pipeline investment that captures future Strategy B clients before any other force does. For the clients above that entry level, the model is clear: instead of acquiring Strategy A firms — expensive, complex, culturally risky — a Strategy B firm can offer its proprietary Agentic AI platform to Strategy A firms on a SaaS or white label basis. The competition becomes a partnership. The threat becomes a distribution network.