Sales tax nexus is the silent killer of solo SaaS — here's the architecture that fixes it

The number nobody quotes

A solo SaaS operator hits $5,000 in monthly recurring revenue. They have customers in 32 US states by month nine. They believe — because every founder forum tells them this — that sales tax is "an enterprise problem" and Stripe Tax will "handle it." Both beliefs are wrong, in a way that produces personal liability under most state responsible-party statutes within 18 months.

The honest version: under *South Dakota v. Wayfair* (2018), every state with an economic nexus rule (currently 45 of the 46 with a sales tax) considers an out-of-state seller obligated to register, collect, and remit sales tax once they cross a threshold. Most thresholds are $100,000 in revenue or 200 separate transactions in a calendar year. A solo SaaS at $5k MRR with a $20 per-month plan crosses the 200-transaction threshold in a single state on roughly the eight-hundredth subscription. They cross it in five states by month four. They cross it in twelve states by month nine. They don't notice because their billing platform doesn't tell them — and the state revenue departments do not send a friendly heads-up; they send a notice of deficiency two years later, retroactive, with penalty and interest.

That deficiency follows the founder personally. State responsible-party rules in roughly 40 jurisdictions pierce the LLC veil for unpaid trust-fund-style taxes, of which sales tax is one. "I didn't know" is not a defense the statutes recognize.

This is the silent killer of solo SaaS, and it is the single biggest reason our spec (section 2.3) treats sales-tax automation as a non-optional v1 capability.

Why Stripe Tax alone is not enough

Stripe Tax is good at one thing: applying the correct rate at checkout once you tell it you have nexus somewhere. It does not track your nexus exposure across all 46 sales-tax states. It does not register you in new states automatically. It does not file your returns. It does not handle exemption certificates. It does not flag when a state changes its taxability rules for digital goods — and roughly twelve states have changed those rules in the last three years, mostly in directions that pull more SaaS into taxable scope.

The vendors that close those gaps in 2026 are Avalara, TaxJar (Stripe-owned, more SMB-skewed), and Anrok (purpose-built for SaaS, newer, narrower). Each one does the four things Stripe Tax does not:

• Nexus monitoring. Every transaction flows through their system; they tell you the day you cross a threshold in a new jurisdiction.
• Registration. They prep and submit the state registration paperwork. You still sign — registration is a Category 5 action under our taxonomy because it creates personal responsible-party liability and you, the human, must consent.
• Returns filing. Monthly or quarterly per state, automatically, with audit-grade workkpapers.
• Exemption certificates. Captured, validated, expired, re-collected. A single unvalidated exemption certificate in a state audit can produce a five-figure assessment on a single transaction.

Our architectural choice is to integrate Anrok as the v1 default for SaaS-shaped tenants (it is purpose-built and meaningfully cheaper at sub-$1M ARR) with Avalara available for tenants that need broader product-tax breadth — physical goods adjacencies, international VAT/GST surfaces — even though physical goods are out of v1 scope for the tenants themselves.

Why the founder is personally on the hook

Two doctrines combine to make sales tax a founder-liability instrument rather than an entity-liability one.

Trust-fund-tax doctrine. Sales tax is money the business collected from the customer *on behalf of* the state. It was never the business's money. Failing to remit it is treated structurally the same way as failing to remit withheld payroll tax — a breach of a fiduciary obligation, not a routine debt of the business. States treat it differently from corporate income tax exactly because of this characterization.

Responsible-party statutes. Roughly 40 states have statutes designating the person with the authority to file returns and remit collected tax as personally liable when the entity fails to do so. The statutory language varies — "responsible person", "person required to collect", "person under a duty to perform" — but the doctrine is consistent: the individual who could have made the entity pay is the individual who *will* pay.

For a single-member LLC, the responsible party is unambiguously the founder. There is no other natural person it could be. The LLC's limited-liability shield does not extend to trust-fund taxes; this is the most-litigated exception to limited liability in US state revenue law, and the state revenue departments win essentially every contested case.

The combined effect: a solo founder who runs a small SaaS through a Delaware LLC, who has not registered in nexus states, who has not collected the tax owed, has a personal balance-sheet exposure that grows by roughly 4-7 percent of revenue per quarter (the blended weighted-average rate across nexus states, before penalty and interest), and that exposure does not go away when the company shuts down.

This is the math nobody puts on the landing page of a "start your SaaS in 5 minutes" product.

The architecture that fixes it

For our own tenants we treat sales-tax compliance as a continuous, monitored, gated workflow — not a year-end accounting task. The shape:

• Day-one Anrok integration. Every tenant has the sales-tax monitor live the moment Stripe Billing is connected. Threshold tracking begins on transaction one. There is no "later we'll worry about that" period.
• Continuous nexus dashboard. The tenant's founder sees a US map. Green states: under threshold and unregistered. Yellow: above 80 percent of threshold (warning). Red: above threshold and unregistered (this is the dangerous state and the system surfaces it as a P0).
• Registration as a Category 5 approval. When a state crosses red, the platform prepares the registration paperwork inside Anrok. The founder receives an Approval Card explaining what triggered the nexus, what the registration commits them to (personal responsible-party designation, monthly or quarterly filings going forward), and a typed-phrase confirmation field. Nothing files until the founder confirms in writing.
• Filings as autonomous Category 3 actions *after* registration. Once a state is registered, monthly returns file automatically through Anrok; the founder receives notification and the workpaper appears in the audit log. This is reversible (an amended return can be filed) and bounded (the amount remitted is exactly the amount collected; we do not advance funds).
• Exemption certificate capture at checkout for any tenant with B2B customers. A blocking modal at the seat-add or invoice-issue step prompts for the certificate. Unvalidated certificates flip the line item to taxable.
• Audit-ready workpapers retained for 7 years, the standard limitations buffer for state sales-tax audits in most jurisdictions.

This is the architecture. It is not novel. Avalara has documented variants for fifteen years. What is novel — and what we believe is genuinely undersupplied in the solo-operator market — is that nobody has packaged this as a default for a $499/month platform user. The vendors price-discriminate toward mid-market and enterprise; the SMB tools wave at the problem and let the founder figure it out.

What we will not pretend

Two honest caveats.

Registration commits you to filings, including months with zero sales. Once a state has you in its registration database, you must file a return every period, even if the return reports zero collected tax. This is a real ongoing cost in attention, even when the underlying tax is zero, and the failure mode (forgetting to file zero returns) produces small but real per-state penalties. Our automation files those zero returns autonomously. A founder who shuts down a tenant must affirmatively deregister — another Category 4 action that surfaces at wind-down.

International is not in v1 scope. UK VAT, EU OSS, Canadian GST/HST/PST, Australian GST are all real, all have nexus or registration thresholds (some as low as zero), and all create personal-officer exposures in their own ways. We refuse non-US-customer tenants at intake until we can do this right. Anrok and Avalara both support it; what we do not yet have is the operational maturity to operate it on a tenant's behalf safely.

Why this is also a moat

The frontier-model providers will not ship this. Sales-tax automation is not a model capability problem. It is a vendor-integration, regulatory-tracking, signing-authority, jurisdiction-by-jurisdiction operational problem. It looks more like a small specialty firm than a software product, which is exactly why incumbents (Avalara, Anrok) command the prices they do.

For an AI-operated company platform, the same logic applies in reverse: the durable differentiation against a future "GPT-X handles your whole company" pitch is the boring, jurisdiction-tracking, signature-routing, audit-evidence-producing compliance layer. Spec section 4.4 makes this explicit and rates our differentiation at 7/10 in part because we have committed to building this layer in v1 rather than waving at it.

If you operate a productized service, a content site, or a small SaaS and you have not yet thought about sales-tax nexus, this is the cheapest piece of advice you will ever take: integrate Anrok or Avalara now, not later. The expensive version of this advice arrives by certified mail from the California Department of Tax and Fee Administration.

Closing

We built the architecture above because the alternative is asking a solo founder to internalize 46 jurisdictions of evolving statute on top of running their company. That is not realistic, and platforms that imply it is are setting their users up.

If you want to see the architecture in motion — including the Anrok integration, the nexus map, and the Approval Card flow for state registration — join the design-partner waitlist below. We onboard ~10 founders per cycle.

Join the waitlist

— The 0h1bai team