Numeral.

A tax company that doesn't feel like one.

On any given Tuesday in San Francisco, a Numeral engineer is watching a state revenue department's portal refresh in a headless browser. Somewhere in Florida, a candle company is shipping its 400th unit of the month and has no idea this is happening. That is the whole pitch. The candle company doesn't need to know.

Numeral does one thing, and it does it for a growing list of brands you have probably bought from: it handles US sales tax and global VAT, end to end, so that founders don't have to. Brex uses it. Character.AI uses it. So does Fishwife, the tinned-fish company that became a minor cultural moment, and 8sleep, which makes mattresses that judge your sleep score.

Two thousand companies, fifty US states, seventy-plus countries, and one stubbornly unsexy problem.

One Supreme Court ruling. Fifty new headaches.

In June 2018, the US Supreme Court decided South Dakota v. Wayfair. The ruling was technical and short on drama. Its consequences were neither. Overnight, every state in the country gained the right to require sales tax from out-of-state sellers - no physical presence required. A small DTC brand selling candles from a kitchen in Vermont could, by Tuesday, owe filings in twenty-two jurisdictions.

The incumbent tax software responded the way incumbent software usually does: with an enterprise sales motion, a six-month implementation timeline, and a calendar invite to talk about your pricing tier. For founders trying to ship product, this was, charitably, suboptimal.

A reminder that the most expensive software in the world is the kind you can't actually start using.

Sales tax was a tax on growth. The faster you sold, the faster you crossed into new states, and the faster you crossed into new states, the faster the legal-and-finance backlog ballooned. Most founders solved this the way humans solve all unpleasant problems: by ignoring it until something broke.

An e-commerce operator and a Stripe engineer walk into a YC batch.

Sam Ross spent four years bootstrapping a portfolio of e-commerce brands to more than $50 million in revenue. He understood, viscerally, that sales tax was the single least pleasant part of running a small online business. He had previously held growth and product roles at Airbnb and Teespring, which is to say he had spent his career around companies whose value scaled with the number of transactions, and whose tax complexity scaled the same way.

Matt DuVall had spent five years as a product engineer at Stripe - back when Stripe was still teaching the internet how to take a credit card - and then led product engineering teams at Notion. He knew what good financial infrastructure felt like from the inside. He also knew what it looked like when it didn't exist yet.

The two joined Y Combinator's Winter 2023 batch and named the company Numeral. The bet was simple, and slightly heretical for the category: build a single product that owns the entire compliance lifecycle, from nexus monitoring to filing to remittance to exemption certificates, and price it so a Shopify store could afford it on day one.

Five minutes a month, or your sales tax back.

The Numeral platform connects to the places commerce actually happens - Shopify, Stripe, Amazon, NetSuite, a few dozen others - and watches the transactions roll in. It tracks where a company is approaching nexus thresholds. It registers the company in new states when those thresholds trip. It pulls the data, prepares the returns, files them, remits the cash, and stores the receipts.

When a state mails a paper notice - and they still do, often - Numeral's virtual mailbox scans it. When a wholesale customer needs to claim an exemption, Numeral collects and validates the certificate. When the rules change in Tennessee on a Tuesday, the customer does not have to learn that the rules have changed in Tennessee on a Tuesday.

The phrase the company keeps coming back to is "less than five minutes a month." It is the kind of claim that should sound like marketing and, talking to customers, mostly doesn't. The whole product is engineered around the idea that the most respectful thing software can do for a founder is leave them alone.

The numbers, which are the point.

Tax compliance is one of the rare software categories where the customer can verify - sometimes painfully - whether the product worked. A missed filing produces a notice. A miscalculated rate produces a refund request. There is no convenient layer of vibes between cause and effect.

So the numbers carry weight here. In its third full year of operation, Numeral has processed more than five billion dollars of transactions on behalf of its customers. It has filed roughly a hundred and fifty thousand returns. Revenue has grown 3.5 times year over year. The company keeps a 4.8-star average on G2, which is the kind of rating you don't get if you accidentally file in the wrong state.

Behind the metrics is a customer list that double-duties as a sort of taste test for early-stage software. Brex, the corporate card platform, runs its sales tax through Numeral. Character.AI does too. So do Ridge, Kitsch, immi, and 8sleep - companies with the kind of transaction volume that makes a tax error very expensive, very quickly.

The investor list reads similarly. Mayfield led the Series B, with Benchmark, Uncork, Y Combinator, and Mantis following on. Total capital raised: $57 million. Valuation, as of last September: $350 million. The company is a little over three years old.

The fastest, easiest way to be boring.

Numeral's stated mission is to make sales tax compliance the easiest and fastest part of running a business. Read carefully, this is a more radical claim than it sounds. The easiest part. Easier than the marketing stack. Easier than payroll. Easier than answering customer support tickets at 11 p.m. The bar is not "no longer painful." The bar is "you forgot it was a thing."

The cultural side of this is harder to photograph but easy to feel in conversation with the team. Numeral is remote-friendly, engineering-led, and aggressive about replacing internal toil with automation before hiring around it. The company is around 120 people. It feels smaller than that, in the good way.

Indirect tax is the next thing software eats.

Payroll got APIs. Payments got APIs. Accounting, slowly, got APIs. Indirect tax - the global, jurisdictional, paperwork-soaked layer of commerce that touches every transaction on the internet - has been waiting its turn. Numeral is the bet that its turn has come.

The category is bigger than it looks. Every US state runs its own rules. The EU is debating real-time VAT reporting. India ships e-invoicing mandates on tight timelines. Brazil exists. Each new jurisdiction is, for a small business, a new reason to either avoid the market entirely or pay an accountant a great deal of money. Or, increasingly, to plug in software that handles it.

If Numeral is right, the next decade of cross-border e-commerce won't look like the last one. Founders won't choose markets based on which tax regime they can survive. They will just sell. The software will figure out the rest.

The candle company never knew.

It's still Tuesday in San Francisco. The Numeral engineer's headless browser has finished its run. The Florida revenue portal has accepted the filing. Somewhere on the East Coast, a candle company is shipping its 401st unit of the month, and a state agency is, quietly, satisfied.

The founder of the candle company is doing something else. She is not thinking about sales tax. She is, if Numeral is doing its job, never going to think about it again.

That is the entire product. That is the entire bet. The boring fintech, doing the boring work, so that the interesting work can happen somewhere else.