The Engine That Finally Broke the Desktop Lock
Luca IQ, a Y Combinator‑backed startup from the S26 cohort, has secured IRS authorization for an API‑first professional tax engine that produces a review‑ready Form 1040 in roughly 30 minutes — the first tax calculator built from scratch in three decades to earn Assurance Testing approval for Form 1040 and operate as an Authorized IRS e‑file Provider (Software Developer, Transmitter, and Intermediate Service Provider).
CEO Angelo Policicchio, a CPA who spent years inside local firms and the Big Four watching preparers retype numbers into software older than they are, founded the company with CTO Akash Sadashivapeth, an engineer who married into a family of accountants and saw every tax tool duct‑taped onto legacy engines from the 1990s. After thousands of conversations with CPAs, enrolled agents, and tax professionals, they rebuilt the foundation rather than layer AI onto existing desktop applications.
The deterministic engine computes returns using rules‑based logic, not probabilistic models. Every identical input produces the identical output, making calculations repeatable, testable, and auditable. Each figure traces back to the specific rule and source document (W‑2s, 1099s, K‑1s, brokerage statements) that produced it. A single API call returns a completed return as structured JSON, a formatted PDF, or IRS MeF XML ready for e‑filing.
That API‑first design inverts the legacy model. Most professional tax software ships as a standalone application, forcing firms to shuttle data between practice‑management systems, document portals, and the tax engine via fragile robotic‑process automations or offshore data‑entry teams. Luca IQ runs underneath the firm's own software as a tax‑calculation service; the firm keeps client records, workflow state, review decisions, and communication. Developers embed the engine directly into proprietary platforms; firms that want a full workflow get client intake, workpaper automation, review, e‑signatures, delivery, and filing in one system. Both paths use the same infrastructure.
Early adopters report saving more than 1,100 hours in a single busy season. Those hours shift toward tax planning and advisory work that commands higher margins. The engine is live, firms are building on it, and the IRS authorization means returns transmit directly without intermediary handoffs. The question now is how fast the rest of the ecosystem responds.
From Data Entry to Review: The 30‑Minute Pipeline
The pipeline moves a client's raw documents (those source forms) into a review‑ready 1040 in about 30 minutes. Firms upload source documents through a client portal that requires no client login. The platform reads each document, maps every value to the exact line on the return it belongs on, and preserves the link so any figure can be traced back to its source page. The calculation engine then runs rules‑based logic so every output ties to a specific rule and the document that triggered it. The result is a completed return delivered as structured data, a formatted PDF, or an e‑filed submission sent straight to the IRS.
The shift is from data entry to review. The platform automates workpaper preparation, converts source documents into structured tax inputs, calculates returns through a deterministic engine, and gives tax professionals an audit trail and final review before filing. The 30‑minute figure is not a fully autonomous filing time; it is the time to produce a return that a CPA can sign off on. The human stays in the loop. What disappears is the grunt work: organizing PDFs, transcribing numbers, cross‑checking schedules against source docs. What remains is professional judgment — evaluating whether a K‑1's passive‑loss limitation applies, spotting a missing 1099‑R, deciding how to treat a crypto transaction the rules engine flags as ambiguous.
The API‑first design reinforces the workflow change. A firm or fintech can call the engine once and get back a finished return in whichever format fits its downstream systems. No separate e‑file step. No re‑keying into a legacy desktop package. Research notes that Luca IQ handles document intake, calculation, and filing in one system rather than routing filings through a separate transmitter or relying on older tax engines beneath newer interfaces. For firms that have spent decades stitching together point solutions, that consolidation is the real time saver.
Early adopters report the practical effect: more returns per preparer, less seasonal hiring, a cleaner path off the desktop stack they've been trying to quit since the 2010s. The company does not publish customer counts or pricing, so independent verification is thin. But the mechanics (document ingestion, deterministic mapping, rules‑based calculation, direct IRS transmission) are specific enough to evaluate. If the pipeline works as described, the bottleneck moves from preparation capacity to review capacity. That shift is what creates the hiring signal the next section examines.
Remote Tax Demand Jumps 135% While Talent Pool Shrinks
Demand for remote tax roles jumped 135 percent between July 2023 and July 2024, while the pool of job seekers grew from 5,400 to 7,300, Distinct Recruitment found in its tracking of the U.S. accounting sector. At the same time, total remote accounting postings barely budged (25,600 versus 25,900), meaning the tax slice is absorbing a disproportionate share of hiring energy.
That energy is colliding with a talent vacuum. The Bureau of Labor Statistics documented a loss of 340,000 accountants between 2019 and 2024. The AICPA recorded a 22.5 percent drop in CPA candidates over roughly the same window, from 95,650 in 2017 to 74,165 in 2024. More than 300,000 U.S. accountants and auditors quit in the past two years alone (a 17 percent decline), and the departures aren't just retiring boomers. Professionals aged 25 to 34 and 45 to 54 are leaving in high numbers, driven by inadequate pay, stalled advancement, and 70‑ to 80‑hour weeks during peak season.
Firms are responding with money and flexibility. ZipRecruiter lists remote tax roles at $111,000 to $163,000 as of June 2026. Tax managers command $133,000 to $207,500, per Cal State Fullerton's business college. Entry‑level candidates now see offers starting at $85,000, well above the $68,000 mean salary for accountants overall. Robert Half reports that many of these roles already offer remote or hybrid arrangements, and the firm's senior leadership expects that percentage to keep climbing as employers adjust recruitment strategies.
| Role | Salary Range (2026) |
|---|---|
| Remote Tax Professional | $111,000 – $163,000 |
| Tax Manager | $133,000 – $207,500 |
| Entry-Level Tax | $85,000+ |
| Accountant Mean | $68,000 |
The remote shift is structural. In 2024, 12 percent of U.S. employees worked mainly from home, triple the pre‑pandemic rate. Finance and accounting sits at 24 percent fully remote, behind only technology (38 percent) and sales and marketing (31 percent), the California Legislative Analyst's Office found. That office also noted employers now draw from a national pool rather than local labor markets, accelerating the geographic decoupling of tax talent.
AI adoption is the catalyst reshaping what those remote hires actually do. Industry surveys indicate growing AI adoption in tax firms, with generative AI tools being piloted for document processing and research. Luca IQ's engine, API‑first, IRS‑authorized, and built to move a client from intake to a review‑ready 1040 in roughly 30 minutes — exemplifies the new stack. Firms adopting it don't need fewer CPAs; they need CPAs who can supervise the engine, validate its output, and handle the exceptions the model flags. That skill set (CPA credential plus AI‑tool fluency) is what the 135 percent demand surge is actually chasing.
The Bureau of Labor Statistics projects 5 percent employment growth for accountants and auditors through 2034, faster than the average for all occupations. But the composition of that growth is shifting. The Big Four shed thousands of roles in 2023, yet layoffs remain rare across the broader profession — "you're a trusted member of the inner circle," ZipRecruiter chief economist Julia Pollak told CNBC. What's expanding is a middle tier: midsize firms, fintechs embedding tax, and specialized boutiques hiring remote specialists who can operate an AI‑augmented workflow from anywhere. The talent shortage isn't solving itself. It's being rewritten.
Incumbents Automate Around the Edges
Luca IQ's IRS‑authorized, API‑first engine landed in a market where the incumbents control the installed base. Drake Software, Intuit, and H&R Block each command millions of filings and decades of preparer loyalty. Their response was not to match Luca IQ's engine (none has announced an IRS‑authorized, API‑native tax calculator built from scratch) but to wrap AI‑driven workflow automation around their existing platforms. The distinction matters: Luca IQ replaces the preparation layer; the incumbents are automating the handoffs around it.
Drake Software moved first in public. On September 16, 2025, the North Carolina‑based company — founded in 1977, serving more than 70,000 firms that collectively prepare over 36 million returns annually, launched Drake Workflow in beta. The suite ties together Drake Tax, Drake Portals, Drake E‑Sign, Drake Pay, and the newly released Drake Tax Online, a cloud‑native, browser‑based preparer tool that starts at $299 for the first user and runs on Chrome, Edge, Firefox, or Safari without installation. Drake Workflow's feature set reads like a project‑management layer purpose‑built for tax season: statuses update automatically as work completes in any Drake tool; tasks route to the right preparer based on return state, with flags for items needing attention; a dashboard shows every client return and its stage; drag‑and‑drop status changes and quick payment tracking replace spreadsheet tracking. Noha Ghazouani, VP of Product Management, said the launch reflects "our commitment to solving the real pain points of today's tax offices," specifically the "intense pressure during filing season" and the need for "real‑time visibility, fewer manual steps, and tools that actually work together." Pricing sits at $299.99 per year, or 40 percent off via the Practice Essentials Bundle.
Intuit's countermove leans on its Google Cloud partnership. The company announced a collaboration that brings Google's Doc AI and Gemini models into TurboTax, expanding done‑for‑you autofill to the ten most common U.S. tax forms. The pitch centers on speed and accuracy for consumers and small businesses, not CPA firms. TurboTax Live Full Service now matches customers to a virtual tax expert or a local Intuit TurboTax Verified Pro for virtual or in‑person preparation. Generative AI innovations roll across TurboTax Online and TurboTax Live, including Spanish‑language translations across the full product line. Intuit Assist, the company's GenAI layer, powers features across both platforms. The architecture remains consumer‑facing; the professional preparer tools (ProSeries, Lacerte, ProConnect) have not received a comparable AI‑native rebuild.
H&R Block, named CNET's Best Tax Service for 2026, added 24/7 AI Tax Assist and free Tax Pro Review to help filers navigate the "One Big Beautiful Bill" changes. The company described a purpose‑specific evaluation framework for its generative AI‑infused solutions: delivery teams partner with front‑line associates to analyze system performance scores across multiple dimensions on every iteration, aiming for a holistic, fit‑for‑purpose view. The assist feature targets DIY filers and the assisted channel, not the professional workflow layer that Drake and Luca IQ address.
None of the three incumbents has claimed IRS authorization for an AI‑first tax engine. Drake Workflow automates handoffs inside the Drake ecosystem. Intuit's GenAI autofill and expert‑matching sit atop the existing TurboTax stack. H&R Block's AI Tax Assist guides consumers and offers a pro review backstop. Luca IQ's differentiation, an API‑first, IRS‑authorized engine delivering that same 30‑minute review‑ready 1040 — remains unmatched in public announcements. The competitive pressure is real, but it is pressing on workflow orchestration, not the calculation core.
The New Hybrid: CPA Credential Plus AI Fluency
The tax preparer who spent 80 percent of a return cycle keying numbers from PDFs into desktop software is not disappearing — but the job description has inverted. Luca IQ's engine converts source documents into structured inputs, runs a deterministic calculation, and surfaces an audit trail for every field. The human now opens a review‑ready 1040 with confidence scores attached to each line item. Their day starts at the validation layer, not the data layer.
Magnetic, which launched an AI agent for 1040 prep in June 2026, makes the same bet differently. Its agent handles contextual analysis such as home‑office deductions, then routes every return to an in‑house enrolled agent or CPA. That reviewer confirms correctness, fixes misses, fills edge cases the system could not resolve, and adds preparer notes the agent did not generate. The reviewer works in an internal tool that surfaces the system's own notes and confidence scores so they know where to look first. In the Spring 2026 season, only 3 percent of Magnetic users claimed the accuracy guarantee (a $10‑or‑more liability change triggers a free return).
The qualifications for this hybrid role are concrete: an active CPA license or enrolled‑agent credential, fluency in at least one AI‑augmented tax platform (Luca IQ, Magnetic, CCH Axcess Expert AI, or Drake's new automation suite), and the ability to trace a calculated field back to its source document in seconds. Firms are also listing "AI workflow configuration" as a desired skill, building review checklists that map to the engine's confidence thresholds, setting escalation rules for low‑score items, and documenting the human‑in‑the‑loop decisions that satisfy audit‑defense requirements.
PwC's January 2026 move to formalize an engineering career track inside the firm signals where the market is heading. The firm separated engineers from technologists because their day‑to‑day work and recruiting pipelines differ. Tax practices are beginning to mirror that split, with some firms creating specialized roles for configuring deterministic engines and managing API connections alongside traditional review roles. Both roles require CPA‑level tax knowledge; the engineer adds Python, API design, and CI/CD for tax logic. The reviewer adds deep familiarity with IRS reference materials and the judgment calls that no model handles — allocation of basis in partial dispositions, reasonable‑cause arguments for penalty abatement, state nexus determinations for remote workers.
Job postings on Indeed for "CPA tax artificial intelligence" roles now include bullet points such as "support and partner with internal IT and digital data teams in integrating tax into enterprise digital and AI transformation initiatives." That language did not exist in 2023 listings. The hybrid specialist is expected to sit in sprint reviews with product teams, define acceptance criteria for new deduction modules, and sign off on regression‑test results when the engine updates its federal‑form library.
Firms that hire for this profile pay a premium. Zero G Talent reported that Stripe's board shows a Business Systems Architect (Tax) role in South San Francisco banded at $274,456–$334,600 — a tax‑engineering hybrid that did not exist as a titled position three years ago. The median salary band across Stripe's 21 salaried roles is $237,000. ASML's median across 56 roles is $152,000, but its tax‑adjacent engineering roles sit at the top of that range.
Tax‑Tech Is Becoming a Primary On‑Ramp for Frontier Talent
The tax‑tech hiring wave sits inside a market that Mordor Intelligence values at $26.5 billion in 2025 and projects to $50.5 billion by 2031, an 11.8 percent CAGR. Coherent Market Insights puts the tax management software segment at $27.8 billion in 2026, climbing to $56.7 billion by 2033 at 10.7 percent. Cloud deployment already carries 60 percent of revenue and runs a 13.6 percent CAGR to 2031. Asia‑Pacific leads regional growth at 15.6 percent. Indirect tax software, driven by e‑invoicing mandates, accelerates at 14.8 percent. These are not niche figures — they are the revenue backbone pulling talent from adjacent frontier stacks.
Venture capital confirms the gravity. AI funding cleared $100 billion in 2024, with generative AI companies alone raising $47.4 billion. Corporate investors supplied roughly a quarter of that capital, seeking strategic alignment alongside returns. The top consumer‑focused tax platform now spends over $100 million per year on foundation‑model partnerships, embedding conversational guidance, audit‑risk scoring, and deduction suggestions across tax, accounting, and payroll lines. Thomson Reuters paid $500 million for SurePrep in January 2023. OpenAI acquired Global Illumination eight months later. M&A is the talent acquisition strategy for incumbents racing to ship generative copilots.
The labor market reflects the pressure. Indeed lists 104,000 Tax AI roles. Frontier‑lab ML engineer compensation has inflated faster than any other tech segment since late 2022, a direct consequence of the GPT‑3.5‑to‑GPT‑4 capability jump that proved commercial viability for LLM‑driven products. Mordor Intelligence flags a shortage of domain‑skilled developers for tax rule engines. Job specs now demand prior experience building AI‑enabled products (particularly RAG systems), experience fine‑tuning base models (ideally via reinforcement fine‑tuning), and a willingness to go deep on tax‑technical problems. That profile blends the frontier‑lab engineer with the CPA who knows the Internal Revenue Code.
First‑party board data from Zero G Talent shows the compensation ceiling. Zero G Talent's data shows machine learning engineers at Stripe range $212,000–$318,000. ASML's median salaried role sits at $152,000 with a band to $235,000. These numbers exceed typical public‑accounting senior‑manager pay, signaling that tax‑tech is bidding for the same talent pool as autonomous vehicles, robotics, and foundation‑model labs.
White‑space niches (microenterprises in emerging economies, cross‑border freelancers, mid‑sized practices needing all‑in‑one collaboration) expand the addressable surface. Vendors now localize via low‑code configurators in under four weeks, bypassing years of traditional development. That speed creates demand for engineers who can ship tax‑logic microservices, not just model researchers.
The 30‑minute return is no longer a demo — it's the new baseline. Every firm that adopts it faces the same choice Luca IQ's early users did: redeploy the saved hours into advisory, or watch the talent that can do advisory walk to the firms that did.
Working in frontier tech? Zero G Talent tracks the openings: see every open ASML role, browse frontier tech jobs, openings at Stripe, and the people building the field.