DOSS is quietly retiring ERP.

A warehouse in Oakland, 7:14 a.m.

A shift supervisor at a fast-growing coffee brand opens his laptop, taps four keys, and watches inventory, purchase orders, freight statuses, and yesterday's gross margin assemble themselves on one screen. He does not call the ERP consultant. He does not file a ticket. He does not, mercifully, open a spreadsheet. He just goes to work.

This is what DOSS sells. Not software, exactly. The absence of friction.

DOSS is a 120-person company in San Francisco. It builds what it calls an Adaptive Resource Platform - ARP, a not-very-subtle dig at the three letters every operator dreads. The platform unifies inventory, procurement, orders, fulfillment, and finance on one real-time data model. It is sold to physical-product businesses: coffee roasters, snack brands, contract manufacturers, the kinds of companies whose problems are decidedly unglamorous and whose software has historically been worse.

ERP has been broken for twenty years. Nobody fixed it.

The unkind truth about enterprise resource planning is that it usually fails. Industry estimates put ERP implementation failure rates between 55% and 75% - a statistic that would end any other software category and yet, somehow, has not ended this one. SAP, Oracle, NetSuite, Microsoft Dynamics - the incumbents charge seven figures, take a year and a half to deploy, and reward the buyer with a system so rigid that changing a tax rule requires a consultant.

Operations teams know this. They paid for it. They live with it. Many of them, given a choice, choose the spreadsheet.

The DOSS founders, Wiley Jones and Arnav Mishra, watched this dynamic up close. Jones came from Athelas. Mishra was a founding engineer at construction-finance startup Siteline. Between them they had spent enough time inside factories in China and high-growth startups in Silicon Valley to notice the same pattern in both: smart operators, smart engineers, brittle software. Hours lost to workarounds. Quarters lost to migrations. Decisions made on stale data because the database refresh ran overnight, when it ran at all.

Their diagnosis was unfashionable. The problem was not the data model. The problem was that the data model was fixed.

What if ERP adapted to you?

Founded in 2022, DOSS started with an opinionated premise: every operation is a snowflake, and any software that pretends otherwise will eventually be the reason someone gets fired. So the founders flipped the contract. Instead of forcing the customer to redesign their business around the software, the software would redesign itself around the customer.

That is the elevator pitch for ARP. Underneath it: a modular system of record, a no-code form and workflow builder, and an AI layer that can model new processes without a consultant chain. It is the kind of architecture that sounds straightforward until you remember nobody has ever shipped it at enterprise scale. There is a reason.

One database. Six modules. Zero consultants required.

DOSS's surface area is broad: inventory management, procurement and purchase orders, order management with 3PL integration, finance and accounting with multi-currency reconciliation, warehouse management, freight and demand planning. Each module pulls from one unified, real-time data model rather than the stitched-together silos that define legacy ERP. The aesthetic is closer to Airtable and Retool than to a 1998 thick client.

The no-code workflow editor is the thing customers tend to mention first. It is the part where the buyer realizes she can change a process without filing a ticket. The AI layer is the part she mentions second. It is the part that drafts new workflows, populates them from existing data, and learns - which sounds modest in 2026 and is, in fact, a small miracle inside this category.

Coffee, snacks, and the supply chain that runs underneath them.

The DOSS customer list reads less like a slide deck and more like a Whole Foods endcap: Kahawa 1893, the women-grown coffee brand; Mezcla, the protein bar with the cult following; Spread The Love, a nut butter house; Noodles.de, a German ramen brand running pan-European fulfillment. These are not pilot customers. These are companies whose entire flow of goods, dollars, and data is now on DOSS.

The investor list reads differently. Madrona and Premji Invest co-led the Series B. Intuit Ventures came in, presumably with a notebook and an opinion. Greyhound Capital and Commerce Ventures joined. Earlier backers - Theory, General Catalyst, Contrary, Mintaka, Pathlight, 47th Street - re-upped. Combined, they have put $73M behind a thesis that has been the graveyard of more than one promising company. The fact that they came back, twice, suggests the unit economics work.

Operations software, but for operators.

DOSS's stated mission - replace legacy, rigid ERP with an adaptive, AI-native platform - is the kind of sentence that has been written before. What is new is the cultural insistence behind it. DOSS does not staff implementation through a partner army. Its own engineers and product experts run the deployments. The cost structure is heretical. The result is that customers go live faster and the team learns where the platform actually breaks. It is a virtuous loop disguised as a labor policy.

The technology stack inside DOSS is, fittingly, modern: TypeScript, React, Next.js, Tailwind on the front; Snowflake and BigQuery for the warehouse layer; Salesforce, Sisense, MicroStrategy, ADAPT, QuickBooks on the integration side. It is the toolkit of a startup that is taking enterprise seriously, not an enterprise pretending to be a startup.

The next decade of physical-goods businesses will run on something. DOSS is betting it is them.

The market is moving. Tariffs, reshoring, demand volatility, AI-driven forecasting - the operating environment for any company that touches a physical good is more complicated than it was three years ago. The software that runs those companies will either adapt with them or it will be ripped out. DOSS is the rip-out.

Whether ARP becomes a category or stays a clever rebrand depends on what every category fight depends on: distribution, retention, the depth of the data moat. The early evidence is encouraging. Customers do not churn off this platform. Investors keep writing larger checks. Competitors are starting to use the word "adaptive" without attribution, which is the kindest form of flattery in enterprise software.

Same warehouse. Same shift. Different software.

At 7:14 a.m., the supervisor in Oakland still opens his laptop. The freight statuses still load. Yesterday's margin still draws itself across the screen. The difference, the one that does not show up on the dashboard, is the one DOSS is selling. He did not, this morning, call anyone for help.

That is the product. That is the thesis. That is, if the founders are right, the new shape of enterprise software.