Evidence-close OS for UK accountancy firms that turns messy sole-trader records into HMRC-ready quarterly filings with reviewer-only exceptions.

1. Mandatory quarterly digital updates have already started, so firms cannot defer fixing the evidence-assembly workflow until some future tax season.
2. The winning product can be automation with sign-off rather than pure self-serve software, because qualified review is still built into the category's trusted offerings.
3. Founders moving from accountant software into direct filing after seeing the same data errors on both sides suggests the real market is the shared records problem underneath both channels.
4. HMRC's widening rollout through 2028 means a firm that nails the first cohort can ride new mandatory cohorts for multiple years without changing the core product.
5. Human-reviewed filing tiers starting at £125 show that buyers already accept paying for accuracy and review, which helps a vertical workflow product monetize early.

Catalyst. HMRC's April 2026 quarterly filing mandate and confirmed rollout through 2028 are forcing a formerly annual workflow into a recurring close motion just as hybrid AI-plus-human review models are proving credible and paid for.

The product plugs into the firm's existing bookkeeping stack and continuously assembles a filing-grade record instead of waiting for a quarter-end scramble. It maps each transaction to a source, flags missing evidence, and groups only the exceptions that matter so reviewers spend time on judgment calls rather than clerical categorization. Each filed number carries provenance back to the underlying bank line, invoice, or payout statement, creating an internal audit trail for sign-off and any later HMRC questions. The first release focuses on firms already serving sole traders at scale, because they feel the regulation immediately and can measure value in reviewer throughput, faster close cycles, and fewer filing corrections.

What's different. Most tax software competes on filing UX or generic bookkeeping automation, then hands the hardest exceptions back to staff in email and spreadsheets. This company starts one layer earlier, at evidence close: it builds a provenance graph for every filed number and optimizes reviewer throughput on the minority of transactions that remain ambiguous. Defensibility compounds through reviewer feedback loops, client-level exception patterns, and integrations into the bookkeeping systems where firms already manage thousands of small-business records.

Jobs to be done

flowchart LR
  Buyer[Accountancy ops lead] --> Pain[Quarterly MTD evidence chaos]
  Pain --> Product[Evidence-close filing OS]
  Product --> Outcome[Faster reviewer-signed HMRC submissions]

Executive takeaways

• MTD for Income Tax creates a real workflow insert, not just a software refresh: HMRC expands the in-scope base from 864k taxpayers in April 2026 to 2.916m by April 2028, with most of the early cohorts already represented by agents [1][2][13].
• The hardest problem is evidence-close, not form submission. ICAEW, ATT, HMRC and vendor materials all point to digital records, quarterly updates, multi-software handoffs and client education as the operational bottlenecks [4][6][7][8][9].
• Incumbents already cover ledgers, receipt capture and end-user filing, so a new entrant does not win by being another filing front end. It must materially improve reviewer throughput on messy source records while fitting into Xero/FreeAgent/QuickBooks-era stacks [14][17][18][19][20][21][29].
• Buyer urgency is high but budgets are constrained. Accountants see MTD as both the biggest challenge and biggest opportunity; they are preparing for more frequent client touchpoints, but they remain sensitive to onboarding effort and thin per-client margins [6][10][11][12].
• Category validation exists for a trust-enhanced model. Record OS raised pre-seed funding around an AI-plus-human-review filing workflow, and Taxfix shows that UK taxpayers already pay meaningful premiums for accountant-reviewed submissions [26][27][28].

Market definition

Category definition: evidence-close workflow software for UK digital-first accountancy and bookkeeping firms handling MTD-for-Income-Tax sole-trader and landlord clients. The job is to convert bank feeds, invoices, receipts and payout records into reviewer-ready quarterly HMRC submissions, while preserving digital links and provenance. Included adjacency: ledger integrations, receipt capture, exception routing, and reviewer sign-off workflows. Excluded: consumer-only tax filing, general ledger systems, and pure document OCR tools with no quarter-close orchestration [1][4][8][14][17][18][19][20][21].

Customer and buyer

Day-to-day users are practice operations leads, bookkeepers and senior reviewers inside online accountancy firms. The economic buyer is usually the practice owner, COO or head of operations because the value proposition is higher reviewer capacity, less junior rework and fewer deadline misses across hundreds or thousands of clients [1][6][9][10][11][32]. Buyers already own ledgers and capture tools, so they prefer overlays that reduce manual review without forcing a stack rip-and-replace [8][14][15][17][18][19][20][21].

Buying triggers

• Mandated quarterly reporting starts with the >£50k cohort in April 2026, then expands materially in 2027 and 2028. [1][2][3]
• Practices are discovering that client segmentation, onboarding, education and team training have to happen before the first filing cycle, not at quarter-end. [6][10][11][12]
• Where firms keep records in mixed ledgers, spreadsheets and capture apps, digital-link and software-compatibility rules create urgency for a more controlled workflow. [4][8][9]

Willingness to pay

Public benchmarks show UK buyers already pay for both core bookkeeping software and trust-enhanced filing help: Xero advertises GBP-denominated monthly plans, FreeAgent publishes paid monthly tiers plus add-ons, Coconut charges subscription pricing for MTD filing, AutoEntry uses consumption-based credits, and Taxfix sells accountant-filed Self Assessment packages starting at £99 [14][16][20][22][28]. That supports a B2B pricing hypothesis of a firm platform fee plus per-active-client-per-quarter charge, especially when framed against reviewer time and avoided rework. [14][16][20][22][28]

Category dynamics

Growth signal ~84% CAGR in mandated taxpayer base from 2026 to 2028 cohorts

Validation signals

• HMRC expects 864k taxpayers in the first 2026 cohort and says 75% of the over-£50k group are already agent represented.
• Accountex found 81.7% of accountants rank MTD as their biggest challenge and 79.08% also see it as a major opportunity, with 34% still unprepared.
• Record OS raised pre-seed capital around AI-prepared, human-reviewed filing, validating investor belief that trust-enhanced tax operations can become a software category.

Regulatory & technical constraints

• Digital record-keeping and digital-link requirements limit manual copy-paste workflows once the first accounting entry has been made.
• Quarterly updates and annual return functionality may live in different software products, raising integration and handoff risk for firms.
• Main-agent versus supporting-agent permissions affect who can maintain records, file updates and submit final returns.
• Quarterly deadlines are fixed and recurring, shifting work from annual peaks to continuous close operations.

Competition is layered. Ledger incumbents such as Xero and QuickBooks own the source-of-truth relationship and HMRC-ready bookkeeping narratives, but they are broad accounting systems rather than exception-first reviewer tools [14][15][29]. Workflow specialists such as Dext and AutoEntry dominate document capture and pre-accounting extraction, yet they stop short of quarter-close orchestration and sign-off routing [18][19][20][21]. Vertical MTD tools such as FreeAgent and Coconut are closer to the filing moment and already speak to sole traders, landlords and accountants, but they are primarily system-of-record or end-user products rather than overlay infrastructure for 1,000-client practices [16][17][22][23][24][31]. The substitute stack today is some mix of Xero/QuickBooks/FreeAgent plus Dext/AutoEntry plus staff review, which means the startup must win on throughput and trust, not basic compliance [6][14][18][20].

Why incumbents do not win by default

• Ledger platforms. Xero- and QuickBooks-style systems do not win by default because they store books, but they do not specialise in evidence gathering, exception queues and reviewer sign-off across fragmented source documents.
• Pre-accounting capture tools. Dext and AutoEntry are strong at extracting receipts and invoices into ledgers, but their natural stopping point is data capture, not filing-grade provenance, reviewer prioritisation or quarter-close controls.
• MTD-ready end-user suites. FreeAgent and Coconut are credible MTD products for small businesses, sole traders and landlords, yet they are primarily destination software. A new overlay can coexist when practices need to keep existing ledgers and only solve exception-heavy close work.
• Practice suites. Capium and similar practice suites promise all-in-one practice operations, but they compete as broad platforms. That leaves room for a thin workflow layer that plugs into the incumbent stack and focuses narrowly on evidence-close throughput.

This company targets UK digital-first accountancy firms that now have to run Making Tax Digital quarterly close for large books of sole-trader clients without proportionally adding qualified reviewers. The pain is not basic HMRC submission; it is rebuilding evidence, chasing missing documents, and resolving ambiguous transactions across Xero, QuickBooks, capture tools, and email before each quarter-end. The proposed wedge is an evidence-close overlay that keeps the incumbent ledger in place, assembles a filing-grade provenance trail, and routes only ambiguous items to a reviewer. This is a narrower and faster-to-prove entry point than launching another consumer filing product or a full practice suite because the buyer already has software but lacks reviewer throughput. The best first proof point is a paid pilot on one reviewer team and one client cohort that measurably increases active clients cleared per qualified tax professional per quarter while reducing close-cycle time and post-filing corrections. Research supports a credible market opening, with a reported $86.0M beachhead SAM and rollout tailwinds through 2028, but exact exception rates by client category, onboarding cost by source-system mix, and customer acquisition cost are still missing. The core strategic bet is that firms will fund an overlay if it fits their existing stack and earns trust through provenance plus reviewer-only exceptions. The biggest disconfirming risk is that incumbents and manual workflows remain "good enough," preventing a separate budget line before the startup proves clear reviewer-capacity gains.

Problem

• Quarterly MTD filing forces UK accountancy firms to repeat low-margin evidence assembly every quarter across hundreds or thousands of sole-trader clients, but the scarce resource is still the qualified reviewer who must sign off.
• Existing ledgers and capture tools handle bookkeeping inputs, yet ambiguous transactions, missing receipts, and fragmented digital-link workflows still spill into spreadsheets, inboxes, and manual rework that erodes margin and creates filing risk.

Solution

• Provide an evidence-close control layer that ingests bank-feed, invoice, receipt, and payout data from the firm's existing stack, links every filed number to source evidence, and produces a reviewer-ready quarterly pack.
• Route only unresolved exceptions to human reviewers with provenance, likely classifications, and missing-evidence prompts so reviewer time shifts from clerical reconstruction to judgment calls and sign-off.

Why we win

• The startup is not trying to replace Xero, QuickBooks, FreeAgent, Dext, or AutoEntry; it sits between those tools and the reviewer, where the research shows the real bottleneck still exists.
• Defensibility compounds from practice-specific exception data, reviewer resolutions, and provenance graphs that improve routing quality and make the product harder to replicate with a generic filing front end.

Milestones

flowchart LR
  Wedge[Reviewer-team pilot in a digital-first accountancy firm] --> MVP[Evidence-close overlay with provenance and exception routing]
  MVP --> Proof[Higher reviewer throughput and faster quarterly close]
  Proof --> Expansion[Full-firm rollout plus landlord, notice-response, and partner channels]

Founding team

Experiment roadmap

Risk assessment

What must be true

• At least half of target digital-first firms experience enough quarterly evidence rework to justify a dedicated software budget in 2026-27.
• The product can raise clients cleared per reviewer by 25% or more without taking over the firm's ledger or filing system.
• Qualified reviewers will sign off from an exception-only queue when provenance is explicit and edit rates are low.
• Xero or QuickBooks-centric deployment covers enough of the beachhead to win the first 10 customers without heavy custom integration.
• Firms that start with one client cohort will expand to broader books or adjacent workflows within 12 months.

Open diligence questions

• How many reviewer hours per quarter do 1000-client digital firms currently spend on evidence chasing and ambiguous categorisation?
• What share of target buyers wants an overlay versus expecting this workflow inside existing tax software?
• Which client cohorts generate the highest exception density and fastest ROI in the first wedge?
• How painful are main-agent and supporting-agent workflow boundaries in real practice operations?
• What implementation burden is acceptable before thin per-client margins erase software ROI?

Model sanity

• Revenue engine. Base-case revenue is driven by growing paying firms from 10 at Y2 exit to 60 at Y3 exit while maintaining a roughly $132K blended customer-year value.
• Must go right. The product must convert paid pilots into production rollouts quickly enough that partner-led distribution can take over from pure founder selling in Y2-Y3.
• Model breaks if. The biggest cash risk is a slower pilot-to-production cycle combined with narrower cohort rollouts, because that cuts both Y3 revenue and the cash cushion in the downside case.
• Next-round proof. The seed case is strongest once the company has 10 production firms, measured 25%+ reviewer-throughput gains, and evidence-request workflows live with buffer cash still on hand.

Scenarios

Sensitivity

flowchart LR
  Leads --> Pilots
  Pilots --> ProductionCustomers
  ProductionCustomers --> Revenue
  Revenue --> GrossProfit
  GrossProfit --> Cash

Flags: The base case still requires a real channel contribution in Y2-Y3; founder-led outbound alone is unlikely to support a 60-firm exit. · Gross margin only stays at 70% if onboarding becomes repeatable on a narrow Xero/QuickBooks-led stack before broader segment expansion. · The model assumes no additional financing after the pre-seed, so any pull-forward into landlord workflows or services-heavy deployments would likely increase the capital need. · Rule of 40 looks unusually strong because Y2 revenue is still small; investors should weight pilot conversion, payback, and implementation repeatability more heavily.

• Reviewer trust gap. If qualified reviewers do not trust the evidence pack, the product becomes another draft generator that still requires full manual checking. Mitigation: Launch with exception-only review, attach provenance to every suggested classification, and prove reduced edit rates on early cohorts.
• Integration sprawl. Firms use inconsistent bookkeeping, document, and bank-feed setups that can make onboarding expensive and brittle. Mitigation: Start with Xero and QuickBooks-centric practices, standardize ingestion around the most common source types, and gate early customers on data hygiene.
• Regulatory edge cases. HMRC rule changes or niche income categories could break automated assumptions and create filing risk. Mitigation: Keep a human-in-the-loop approval step, maintain rules and exception libraries by cohort, and expand coverage category by category instead of claiming full automation too early.