Finta's Two-Person Team Files Taxes for 600 Startups. Bookkeepers Charge $700 a Month.

Why This Story Covers Venture-Backed Startups, Not Everyone

Finta, a two-person AI accounting platform that launched from Y Combinator in 2023, has reached 600 paying customers and profitability without a sales team. Its founder writes the code and holds the tax license that signs every return. The platform categorizes 96 percent of transactions automatically and files a tax return in about thirty minutes.

AI-native accounting platforms like Finta are automating bookkeeping and tax filing for venture-backed startups, triggering rapid market growth and forcing legacy players, regulators, and the accounting workforce to adapt. The beachhead is narrow: U.S. Delaware C-Corps, venture-funded, 5 to 200 employees, bank and card connections at Mercury or Brex or Ramp. But the ripple effects are already hitting QuickBooks, the IRS, and a labor market splitting in two.

The break isn't happening at the local coffee shop, the regional law firm, or the mid-market manufacturer running on NetSuite. It's concentrated in a high-velocity slice of the economy: venture-backed, Delaware C-Corp startups that raise priced rounds, hire fast, and operate with lean or nonexistent finance teams. These companies share a uniform legal structure, a common cap-table rhythm, and investor pressure to produce clean books for the next board meeting or due-diligence request. They also share a problem — traditional bookkeeping costs $700 a month and up, scales with expense volume, and still requires a founder to click "Categorize" hundreds of times. That friction created the opening AI-native platforms are now exploiting.

Since 2020, the U.S. has averaged more than 5 million new business applications per year, the Census Bureau's Business Formation Statistics show. A disproportionate share of the venture-backed cohort lands in the Bay Area, where 60-plus AI accounting and tax startups have raised capital since 2023 — Bay Area firms alone account for $1.2 billion-plus, roughly 30 percent of all deals in the category. At the same time, the accounting profession has shed more than 300,000 practitioners since 2020, the Wall Street Journal reports, and 75 percent of U.S. CPAs are projected to retire in the next decade. Firms cannot hire their way out of the backlog; they buy software instead.

We are not covering the 30 million U.S. small businesses that run on QuickBooks and a part-time bookkeeper. We are not covering enterprise finance teams that have already standardized on Workday or SAP. We are not covering non-U.S. entities, non-C-Corp structures, or companies with complex multi-entity consolidations that require human judgment at every layer. The rest of the article follows the ripple.

Finta's Trajectory: From YC Seed to 600 Customers and Profitable

Finta launched in 2023 from the Y Combinator Winter 2023 batch with a $2 million seed round backed by YC itself, Mercury CEO Immad Akhund, and operators from Mercury, Brex, and Ramp, plus Andreessen Horowitz, Costanoa Ventures, Forerunner Ventures, LGF, Pioneer Fund, and a handful of angels. The cap table reads like a map of the modern fintech stack. Two years later the company reports 600-plus paying customers, revenue that has tripled each year, and profitability. All of it came through word of mouth. No sales team. No marketing spend. According to Zero G Talent's board data, the first outbound hire, a Founding Account Executive at $210K–$240K, was posted only recently as Finta begins building a go-to-market motion from scratch.

The founder, Andy Wang, is the sole technical founder. He writes the code and holds that license, signing every return Finta files. The core team totals two people. That headcount is deliberate: Wang has said the goal is to replace the bookkeeper and QuickBooks with software that bundles bookkeeping, tax preparation (federal, state, franchise, R&D credits), and a real-time dashboard for burn, runway, and MRR into a single subscription. Taxes and R&D credits attach to every account; the accounting subscription is the anchor.

Early customers include Resend, Onyx, Promptfoo, and Skyvern — names that signal the product has penetrated the YC-adjacent founder network. The pitch is straightforward: traditional bookkeepers charge $700-plus per month with opaque pricing that scales with expenses; QuickBooks demands hundreds of manual categorization clicks each month; Google Sheets breaks at the first audit. Finta's alternative is a flat, transparent price, a ten-minute onboarding that connects to Brex and Stripe, and a free month-end close trial so prospects can compare output side by side with their current setup.

The unit economics work because the automation rate is high enough to keep marginal cost near zero. Wang has argued that no VC or customer ever requires a startup to be on QuickBooks (in fact, no one likes QuickBooks) and that the risk of leaving it is negligible. The competitive set is legacy bookkeeping services like Pilot (which still operate inside QuickBooks) and the incumbent itself. Finta's differentiation is the tax license: the same platform that categorizes transactions also signs the return. That closes the loop that usually requires a separate CPA firm.

Now the company is hiring a Partnerships Lead at $165K–$190K and a GTM Lead (Events & Partnerships) at $120K–$160K, alongside earlier technical roles: Founding Engineer ($150K–$250K), Founding Design Engineer ($100K–$180K), Founding Designer ($140K–$220K). Zero G Talent's figures put the board at six salaried roles with a median band of $205K. The shift from pure product velocity to commercial execution is the inflection point every AI-native accounting platform will face: the technology proves itself, then the distribution problem arrives. Finta is living through it in real time.

The Market Behind the Momentum: AI Accounting Adoption Data

The numbers moved faster than the profession expected. Wolters Kluwer's 2025 Future Ready Accountant Report found AI adoption inside accounting firms jumped from 9 percent in 2024 to 41 percent in 2025 — the fastest single-year acceleration measured in any segment of the profession. Gartner's finance function survey tracked a parallel climb: 37 percent in 2023, 58 percent in 2024, and 59 percent in 2025. The gap between finance and other administrative functions (HR, legal, procurement) that Gartner measured at 2x in 2023 had nearly vanished by 2024.

Market sizing reflects the same velocity. The AI-in-accounting market sat at roughly $6.93 billion in 2025 and is projected to reach $10.4 billion in 2026, a 50 percent compound annual growth rate per The Business Research Company. Mordor Intelligence puts the 2026 figure at $10.87 billion, growing to $68.75 billion by 2031 at 44.6 percent CAGR. Scott's International estimates a $37.6 billion market by 2030. The spread across forecasts is wide, typical for a category in breakout, but the direction is not.

Where the technology is actually landing tells a more specific story. Gartner's November 2025 survey of finance functions using AI found three use cases above one-third adoption: knowledge management at 49 percent, accounts payable automation at 37 percent, and error and anomaly detection at 34 percent. A year earlier the same survey highlighted intelligent process automation (44 percent), anomaly detection (39 percent), analytics (28 percent), and operational assistance (27 percent). KPMG's 2024 U.S. finance report showed stark variation by sub-function: financial planning 78 percent, accounting workflows 76 percent, treasury management 64 percent, financial reporting 52 percent. U.S. firms told Wolters Kluwer their most common uses were tax, accounting, and audit research (42 percent), productivity tooling (41 percent), and document summarization (38 percent).

The ROI data is why budgets are moving. KPMG reported 92 percent of companies say finance AI initiatives meet or exceed ROI expectations; among AI leaders that figure rises to 96 percent, with 61 percent reporting higher-than-expected returns. Wolters Kluwer found 83 percent of firms globally posted revenue growth in 2025 (up from 72 percent in 2024) and 79 percent saw improved profitability. MIT and Stanford researchers measured the operational delta: accountants using AI support 55 percent more clients per week and cut monthly financial close by 7.5 days. London School of Economics put the time savings at one full workday per week for frequent users.

Yet the adoption curve has a kink. Gartner's latest survey shows momentum slowing, 58 percent to 59 percent year over year, with two drag factors: 16 percent of finance organizations report no planned AI implementations for the coming year, and 25 percent remain uncertain how to move from planning to piloting. CPA.com and AICPA found 63 percent of firms "exploring" AI but only 16 percent fully implemented in operational workflows. Wolters Kluwer's May 2025 survey of 392 finance leaders placed 86 percent of North American teams in early stages: 53 percent exploring, 33 percent piloting, just 4 percent scaled across multiple functions. The skills gap is concrete — 56 percent of accounting professionals cite generative AI as their largest skills shortage area per CPA.com, and 31 percent of firms globally flag advanced technical skills as a top staffing challenge, up from 27 percent a year ago.

Generative AI adoption is further along than the aggregate numbers suggest. KPMG found 58 percent of U.S. finance functions piloting or deploying gen AI in 2024. Intuit QuickBooks reported 64 percent of accounting firms plan to invest in or upgrade AI in the next year, up from 48 percent in 2023. Forty-six percent of accounting professionals now use AI daily — nearly double the 28 percent rate among small business owners.

The next inflection is agentic. Wolters Kluwer projects agentic AI adoption in finance teams will hit 44 percent by end of 2026, a 600 percent increase from the current 6 percent. Gartner's data shows 40 percent-plus of enterprise accounting applications will embed AI agents by 2026. Avalara has already shipped AI tax agents that file returns 24/7 without human input; Thomson Reuters released "Ready to Review," automated 1040 preparation that generates draft returns for accountant verification rather than accountant preparation. The shift moves the accountant from preparer to reviewer — a leverage ratio change that compounds the economics driving platforms like Finta into the infrastructure layer.

Seventy-seven percent of firms plan to increase AI investment over the next three years; among high-growth firms the figure is 87 percent. The capital is committed. The workflow redesign is underway. The only open question is how fast the remaining 86 percent of North American finance teams move from pilot to production.

The Competitive Response: QuickBooks, Ramp, and the Fintech Platforms

Intuit saw the threat early. The company introduced Intuit Assist in September 2023 on its GenOS platform, then made it generally available in QuickBooks Online for U.S. users on November 20, 2024. That was the warm-up. The real move came July 1, 2025: Intuit pushed a "virtual team" of AI agents to every QuickBooks Online user, including categorization, invoice chasing, a light CRM, and payroll collection, with no opt-out. The agents arrived the same day prices jumped across every tier.

QuickBooks community threads called the increases "unacceptable" — one user described a roughly 70 percent hike for the same basic features. Intuit's headline claims: AI-drafted invoice reminders get businesses paid 45 percent faster, about five days sooner, and the agents save up to 12 hours a month. The professional consensus is blunter. School of Bookkeeping's guidance: "trust, but verify — especially when closing books or posting in bulk." Users report spending time every week fixing miscategorized transactions. Even Intuit's own guide for accounting pros tells them to review the agents' work.

The incumbent is also buying distribution. Intuit signed a multi-year contract worth more than $100 million with OpenAI in November 2025. By March 5, 2026, the QuickBooks app was live in ChatGPT for logged-in Free, Plus, and Pro users in the United States. The deal gives Intuit access to ChatGPT's audience and deepens its use of OpenAI's frontier models across its platform. Intuit also released a local MCP server as an early developer preview in October 2025 (sandbox-only, OAuth through the developer portal), but outside AI can still only read QuickBooks, not act on it.

Ramp took a different angle. On June 3, 2026, the company launched Ramp Stack, an AI operating system built specifically for accounting firms. Stack deploys agents for month-end close, reconciliations, and journal entries. It connects to the systems where financial data lives and completes accounting work for teams to review and approve.

Brex entered the same week with its AI-native Accounting API, delivering real-time, two-way ERP integrations that automate workflows from transaction to close. SeatGeek, a Brex customer, achieved a 99 percent employee compliance rate and accelerated documentation processes tenfold. Brex's approach targets the embedded layer, letting product teams weave bookkeeping, reconciliation, and reporting directly into their apps.

Anthropic arrived in May 2026 with Claude for Small Business, integrating Claude Cowork with QuickBooks, Canva, DocuSign, HubSpot, and PayPal. The company plans a coast-to-coast promotional tour starting in Chicago. OpenAI's Enterprise ChatGPT, launched in late 2023 with ChatGPT Business for smaller teams, already had a head start.

Intuit's response to all this pressure included cutting 3,000 roles, 17 percent of its workforce, in May 2026. CEO Sasan Goodarzi said the layoffs would reduce complexity and focus resources on AI. Revenue in the fiscal second quarter ended January hit $4.65 billion, up 17 percent, with net profit of $693 million, up 48 percent. Yet Intuit's shares have consistently underperformed the S&P 500 over the past 12 months, caught in the broader worry that traditional SaaS firms cannot keep up with AI-native products.

Legacy platforms are bolting AI onto existing workflows and gating it by tier. Fintech infrastructure companies are building AI-first operating systems for the accounting layer itself. The next battleground is whether embedded accounting APIs from Brex, Ramp, or newcomers can displace the general ledger as the system of record.

The Regulatory Horizon: IRS Guidance on AI Tax Preparation

The IRS moved in June 2026. The Office of Professional Responsibility, the unit that enforces Treasury Department Circular 230, issued its first formal interpretation of how those rules apply when a tax practitioner uses generative AI. The alert, published as Issue Number 2026-19, does not create new duties. It maps existing ones onto a technology that can hallucinate case citations, leak client data across engagements, and compress ten hours of research into thirty seconds of compute.

Circular 230 is the federal regulation that governs anyone who practices before the IRS: CPAs, attorneys, enrolled agents. The OPR has teeth — censure, suspension, disbarment. Earlier in June the IRS merged OPR with the Return Preparer Office into a new Tax Professional Management Office, signaling that oversight of AI-assisted work will be centralized and sustained.

The alert covers six specific sections of Circular 230. Due diligence in preparing and filing documents (§ 10.22). Competence and technological literacy (§ 10.35). Firm procedures and supervisory responsibilities (§ 10.36). Requirements for written advice (§ 10.37). Fee practices and client billing (§ 10.27). Protection of taxpayer data under IRC §§ 6713 and 7216(a). Each section gets a worked example of how AI can trip a practitioner.

The central message is clear: AI may assist a practitioner, but it cannot replace the practitioner's professional judgment or serve as the practitioner of record.

That line, from the JDSupra analysis of the alert, captures the through-line. Practitioners must independently review and verify every fact, citation, and calculation an AI system produces before submitting it to the IRS or delivering it to a client. The OPR warns that courts have already sanctioned lawyers for filing GAI-generated briefs peppered with fake citations — financial penalties, public censure, ethics courses, default judgments, disciplinary referrals. The alert explicitly flags a Deloitte Australia episode where an AI-produced government report contained invented quotes and references to non-existent books, resulting in a partial refund.

Confidentiality is the sleeper risk. The OPR notes that virtually all professional tax firms already use some form of AI, often embedded in research platforms like Westlaw Edge or Bloomberg Tax. But generative AI changes the data flow. Client information uploaded to a public or unsecured model can be repurposed to answer another user's query, or spill across matters inside the same algorithm. The alert mandates secure, enterprise-approved systems only. Willful mishandling of taxpayer data through AI triggers Circular 230 discipline.

Billing gets its own spotlight. Section 10.27 prohibits unconscionable fees. The OPR calls out double-billing, charging for AI-assisted work and then billing again as if the task were done manually. A noticeable pattern of billing differentials across clients may constitute a violation. Cost savings from AI should be passed on openly, with billing practices that reflect actual efficiencies.

Firm leaders cannot claim ignorance. If staff use consumer AI tools without oversight, that is a firm-level compliance failure under § 10.36. The alert lays out a pre-filing-season checklist: identify approved AI tools and permitted use cases; prohibit entry of client data into public platforms; train personnel on risks, confidentiality, and verification obligations; establish review procedures for AI-assisted filings and written advice; document AI use and verification steps; evaluate third-party providers before permitting their use; monitor tools for accuracy, security, and compliance.

States are moving in parallel. California, Colorado, Illinois, and Utah have enacted AI governance legislation focused on transparency, bias reduction, and consumer protection. The American Bar Association issued Formal Opinion 512 on generative AI tools in July 2024. The regulatory surface is expanding, not contracting.

For AI-native platforms like Finta, which automates bookkeeping and tax filing for startups that are venture-backed, the OPR alert is a design constraint, not a suggestion. The platform's 96% transaction categorization accuracy and "squeaky clean books" promise must be backed by human review workflows that satisfy § 10.22 due diligence and § 10.35 competence. The founders (Andy Wang, Pavel Asanov, Kevin Siskar, Ryan F. Salerno) are hiring a Founding Account Executive and a Partnerships Lead in San Francisco, roles that will need to articulate how Finta's architecture keeps client data off public models and produces an audit trail the OPR would accept.

The practitioner remains responsible for the final filing, regardless of how much underlying work an AI system performed. That sentence, repeated across the OPR guidance, the Thomson Reuters analysis, and the JDSupra summary, is the regulatory floor. Every AI-native tax product now has to build above it.

The Labor Market Ripples: What Happens to Bookkeepers and Accountants

BLS labor statistics draw a clean line between the two halves of the accounting workforce, and the line is moving. BLS labor statistics show employment for bookkeeping, accounting, and auditing clerks at 1.51 million workers (SOC 43-3031) with a median $49,210 as of May 2024, projected to decline 6 percent between 2024 and 2034. Over the same decade, BLS labor statistics show employment for accountants and auditors at 1.42 million licensed professionals (SOC 13-2011) earning a median $79,880 as of May 2025, projected to grow 5 percent. The total accounting workforce exceeds 2.9 million, but the split is widening.

The wage gap, roughly $30,000, is not new. What is new is the speed at which the mechanical half of the clerk's job is being compressed. A QuickBooks survey put the figure higher: 98 percent of accountants and bookkeepers have already used AI to help a client, and nearly half report using it daily. The tasks moving first are exactly the ones that used to fill a clerk's day: transaction coding and expense categorization, data entry from receipts and invoices, invoice processing (matching purchase orders to bills to payments), and basic reconciliation. These are high-volume, pattern-based, low-ambiguity tasks — precisely what large language models and OCR-based tools handle well.

The BLS numbers aren't predicting the death of bookkeeping as a profession — they're predicting it will die as pure data entry.

BLS projection methods assume technological change proceeds at a pace consistent with past experience. The agency's own researchers note that while digital cameras forced a model adjustment for photographic process workers in the early 2000s, no such adjustment was made for truck drivers despite autonomous-vehicle speculation in the 2010s. The current projections for clerks and accountants were built before the adoption spike of 2024–2025. Each year BLS issues new 10-year projections incorporating new data, but the lag is real.

The hiring data already shows the leading edge. Census Bureau economist Lee Tucker found that hiring of workers aged 22–24 dropped 9 percent immediately after ChatGPT launched in late 2022 for AI-exposed industries (finance, insurance, professional services) compared with all other industries. Between Q3 2022 and Q2 2025, employment for early-career workers in those industries fell 12–15 percent, roughly 150,000 fewer jobs. Stanford's Digital Economy Lab had earlier documented 16 percent slower employment growth for early-career workers in AI-exposed roles (software developers, marketing professionals, sales managers) between mid-2024 and September 2025. The unemployment rate for bachelor's-degree holders 25 and older jumped to 3.2 percent in August 2025 — the first time above 3 percent in roughly nine years outside the pandemic. White-collar unemployment has risen each year since 2023; blue-collar unemployment has largely declined or held steady.

The displacement is not uniform. Clerical and administrative roles are among the first to be automated. Credit analysts' employment is projected to decline 3.9 percent from 2023 to 2033. Yet personal financial advisors are projected to grow 17.1 percent over the same period — demand for human counsel in complex financial decisions persists, particularly for older clients. The pattern inside accounting mirrors this: the work disappearing is repetitive and rules-based; the work expanding is judgment-based.

What remains stubbornly human are the tasks requiring a judgment call rather than a pattern match: deciding whether a borderline expense counts as marketing or professional development; building depreciation schedules and amortizing prepaid expenses; estimating accrued liabilities and inventory write-downs; interpreting an unusual transaction that doesn't fit a template; explaining why the numbers look the way they do to a nervous business owner; taking on the compliance responsibility for accuracy under GAAP and IRS rules. An AI tool can flag an anomaly, but it cannot sign off on a set of books or take responsibility if something is wrong. Someone with professional judgment and legal accountability still has to own that call.

That accountability bottleneck is visible in the hiring data. Ninety percent of finance leaders report they cannot find enough qualified accounting professionals, even as entry-level data-entry roles shrink. The constraint has shifted from "who can enter data" to "who can supervise the AI, catch what it misses, and advise the client." Workers with AI skills in business and finance roles command a 56 percent wage premium over peers without them. Dedicated AI-accounting specialist roles have grown 26 percent with salary premiums of $15,000–$25,000.

The relationship is changing. Less "did you enter last month's transactions," more "here's what the pattern in your data means for your cash flow." DIY bookkeeping becomes more viable for very small businesses and solo freelancers (one revenue stream, a handful of recurring expenses), but the caveat holds: an AI tool can suggest a categorization; it cannot tell you whether that categorization is defensible if the IRS asks. Inventory, multiple entities, contractors, payroll, accrual accounting — the judgment calls pile up fast enough that professional oversight still pays for itself.

The cheapest possible bookkeeping service, built entirely around manual transaction entry, faces real pricing pressure as AI-assisted competitors do that same work faster. The bookkeeping relationship worth paying for is the one where a human reviews what the AI produces, catches what it misses, and turns your numbers into decisions you can act on. Watch the next BLS projection cycle; it will tell you whether the 6 percent decline estimate holds or whether the adoption spike forces a revision.

What to Watch Next: Embedded Accounting, AI Tax Filing, and the Next Frontiers

The quietest bet in the AccountingTech Startup 100 might be the most consequential. Embedded accounting infrastructure, just two companies in the current taxonomy, represents the thesis that bookkeeping will vanish into the software SMBs already use. Layer powers 30,000-plus SMBs through APIs that sit inside vertical platforms; its $6.6 million seed came from Emergence. Open Ledger, with a $10.5 million seed, targets SaaS and banking-as-a-service platforms with an AI engine that handles reconciliation and reporting natively. The pattern mirrors embedded finance: Swan, which secured €40 million in 2023, processes over €7 billion in transactions for 100 customers by letting non-banks offer banking without becoming banks. Fragment, which raised $9 million in 2024, gives fintech developers a ledgering API that keeps product balances in lockstep with bank statements — the reconciliation layer that prevented the Evolve Bank and Synapse meltdown. Brex took the same logic to the ERP layer in January 2026, launching an AI Accounting API that syncs transaction data continuously into Rillet and Campfire, two AI-native ERPs that have raised roughly $100 million and $65 million respectively. The manual month-end close, Rillet's CEO Nicolas Kopp said, is a relic. Brex's API already eliminated 10,000-plus hours of manual work in its first months.

Tax filing is the next surface for embedding. April, an embedded AI tax-filing engine that fintechs run inside their own apps, closed a $38 million Series B led by QED at roughly $78 million total raised. Around it clusters a cohort of AI tax specialists: Black Ore builds an enterprise platform for the tax profession; TaxGPT positions itself as an AI operating system for research, prep, and review; Juno, CPA-founded, automates SMB firm returns on a $12 million seed; Filed raised $17.2 million for its AI tax preparer; Magnetic, a Y Combinator company, targets CPA firms; Instead (formerly Corvee) offers 1,500-plus AI-driven tax planning strategies with a strategic investment from IRIS Software Group; Accordance, backed by Khosla and General Catalyst, deploys a multi-agent "AI brain" for tax, audit, and CPA teams; Truss, an AICPA/CPA.com 2026 accelerator graduate, covers the end-to-end workflow from intake to delivery. The IRS has accelerated the timeline. Its first formal guidance on AI applies Circular 230 professional standards to tax practitioners using AI tools. Meanwhile, IRS Direct File, the government's free-filing experiment, has been formally discontinued for the 2026 season. The agency is rolling out its own AI-powered modernization that will change how small businesses file in 2026. The vacuum left by Direct File and the new compliance burden on AI-assisted preparers create a clear opening for embedded tax engines that sit inside payroll, banking, and accounting platforms.

A third frontier skips software sales entirely. Oath, a CPA-licensed AI audit firm, aims to automate roughly 80 percent of audit work; its ~$6.6 million seed includes a former PCAOB board member. Cranston, a Y Combinator-backed full-stack AI accounting firm, puts licensed CPAs in the review loop. Rational, which raised $500 thousand in September 2026, and SoVa, at $2.7 million the same month, join the "replace the incumbent" model — not selling tools to firms, but building the firm itself around AI. Thrive Holdings is executing the roll-up version: raising roughly $2 billion to buy accounting firms and rewire them with AI; its arm Current has already acquired 48 firms. The buyers now have an AI thesis, not just a multiple. Bench, the tech-enabled bookkeeping pioneer, was acquired by Employer.com in 2026. Brex, which built the AI Accounting API now embedding into ERPs, was acquired by Capital One for roughly $5.15 billion the same year. The M&A signal is unambiguous: distribution and trust moats belong to the firms, and AI is the lever to unlock their margins.

Capital flow confirms the vector. VC-backed accounting startups total 377 companies with $2.7 billion in aggregate funding, averaging $26.5 million per company. Seed rounds land in the $2–6 million range; Series A rounds hit $8–20 million; category leaders command far more. The hottest sub-sectors: AI-native ledgers and continuous close, accounts payable and receivable automation, spend and expense management, sales-tax and e-invoicing compliance, plus embedded accounting APIs, audit automation, and outsourced bookkeeping services. Geographic concentration remains San Francisco Bay Area and New York, with London, Toronto, Austin, and Salt Lake City growing; engineering teams frequently sit in India. Compliance-focused startups cluster near the tax regimes they serve. Global e-invoicing mandates and real-time tax reporting requirements are turning compliance from a cost center into a product category. Thomson Reuters found the top five generative AI applications in accounting are tax returns, tax research, tax advisory, bookkeeping, and document summarization — exactly the workflows these new platforms target.

The open question is whether the embedded layer stays horizontal or fractures by vertical. Layer and Open Ledger bet horizontal. April bets on fintech distribution. Oath and Cranston bet on the CPA license as the ultimate moat. Thrive bets on roll-up economics. Finta, profitable at 600-plus customers, bets on the venture-backed startup wedge. Wang still writes the code and signs the returns. Two people, 600 companies, and the books close in thirty minutes. The beachhead holds — for now.

Working in AI? Zero G Talent tracks the openings: see every open Finta role, browse AI jobs, the companies hiring, and the people building the field.