Every software company with an AI strategy is now claiming to be AI-native. Most are bolting a chatbot onto a product designed in a pre-AI era and calling it transformation. The architecture underneath hasn't changed. The pitch deck just got a new slide.

A Stockholm-listed company is making a louder claim, and it just put the first piece of evidence on the table for anyone to use.

Done.ai, traded on Nasdaq Stockholm, has opened Done CRM to all customers, the first native application within Done OS, its AI-powered operating system for small and medium-sized enterprises. After a waitlist phase that started in April, the CRM is now open for any business to sign up. The company frames it not as a CRM with AI features stapled on, but as the opening move in a much larger platform play.

The distinction Done.ai is drawing is worth unpacking, because it's the whole pitch.

Traditional CRMs are systems of record. You put customer data in, the data sits there, and when you need it you go look it up. The intelligence, such as it is, lives in the human staring at the screen trying to figure out what to do next. Done.ai's argument is that this model is obsolete. Done CRM is built for AI from the foundation, with an architecture meant to understand context, surface insights, and act on data in ways a conventional relational database simply can't.

What does that look like in practice? Because Done CRM lives inside Done OS, it can read invoice history, account balances, and payment behavior directly from the ERP data on the same platform. That lets it trigger upsell signals, churn alerts, and next-best-action recommendations without a salesperson hopping between five disconnected systems to assemble the picture manually. The CRM doesn't just store the relationship. It watches it and tells you when to move.

The CRM is the headline. Done OS is the strategy.

Done.ai's wager is that small and medium businesses are drowning in disconnected tools. A CRM here, an accounting package there, a separate billing system, a project tracker, all of them holding fragments of the same customer relationship and none of them talking to each other. The promise of an operating system is that everything a business needs sits connected on a single intelligent layer, where an AI can actually reason across the whole picture instead of one silo at a time.

That's a genuinely hard thing to build, and it's why the CRM matters as a proof point rather than a product in isolation. Done OS continues its rollout with more native applications planned through 2026. The CRM is simply the first one out the door, the test of whether the architecture delivers on the single-layer promise. If it does, each additional app gets more valuable because it can draw on everything the others know. That's the compounding logic behind any platform play, and it's the same logic driving the broader push toward AI that moves from passive tool to active operator.

The financials give the strategy some weight. Done.ai reported Q1 2026 net revenue of 139.3 million Swedish kronor with positive EBITDA, which is more than most companies talking this ambitiously can claim. A public listing also means the numbers get scrutinized every quarter, so the platform story has to translate into revenue, not just vision. That discipline can be a feature. It forces the company to ship things people actually pay for.

SME software is one of the most contested markets in all of tech. The incumbents are entrenched, the switching costs are real, and small businesses are famously reluctant to rip out tools they already know. Done.ai is betting it can leapfrog the field by offering something the incumbents structurally can't: a unified, AI-native layer rather than a patchwork of acquired products duct-taped together.

The challenge is that incumbents have distribution, brand trust, and years of accumulated customer data. Done.ai has architecture and timing. Whether clean design beats entrenched distribution is the oldest question in enterprise software, and the answer usually comes down to whether the new thing is enough better to overcome the inertia. An AI layer that genuinely saves SMEs hours every week might just clear that bar. A marginally nicer CRM won't.

The phrase gets thrown around so loosely that it's almost lost meaning, so it's fair to ask what Done.ai actually means by it. The company's claim is architectural, not cosmetic. A bolt-on AI feature sits on top of a relational database that was never designed to be reasoned over by a model. You can ask it questions, but the underlying structure fights you, because it was built to store rows, not to surface context.

Building for AI from the foundation means the data model itself is designed so an AI can understand relationships, infer intent, and act. That's a much heavier lift, and it's the kind of thing that's nearly impossible to retrofit onto a legacy product without rebuilding it. If Done.ai has genuinely done that, the moat isn't the CRM. It's the architecture that lets every future app be smarter than a bolt-on competitor's equivalent. If it hasn't, the AI-native label is just another slide, and the market will figure that out fast.

One more thing worth saying plainly. Small and medium businesses are notoriously hard to sell to. They have no IT department, little patience for migrations, and a deep, rational suspicion of any vendor promising to replace the five tools they already grudgingly tolerate. Winning them takes more than a better architecture. It takes onboarding so smooth that switching feels easier than staying.

Done.ai's controlled rollout from a waitlist, rather than a splashy open launch, suggests the company knows this. It onboarded customers gradually to protect product quality and tune the experience before opening the doors. That's the right instinct for this buyer. An SME that has one bad week with a new system doesn't file a support ticket. It quietly goes back to the spreadsheet and never returns.

Being listed on Nasdaq Stockholm shapes how this story plays out. A public company can't quietly miss a roadmap and spin it in the next all-hands. Every quarter, the platform ambitions get measured against actual revenue, and the market votes with the share price. That pressure is a double-edged thing. It forces discipline, but it also rewards visible progress over patient platform-building, and operating systems are nothing if not patient projects.

Done.ai's positive EBITDA gives it some breathing room that a cash-burning private competitor wouldn't have. It isn't racing a fundraising clock to prove the Done OS thesis. It can fund the rollout partly from operations, ship apps at a sustainable pace, and let the compounding logic do its work. For a platform bet, that financial footing might matter more than any single feature. The companies that win these long games are usually the ones that don't run out of money before the strategy pays off.

Opening the CRM to everyone is a milestone, not a verdict. The real test of the Done OS thesis comes with the next native application and the one after that.

A single app, however well built, doesn't prove the operating-system claim. The magic only shows up when multiple apps share the same intelligent layer and start doing things no standalone tool could. When the CRM knows what the billing system knows and the project tracker knows what the CRM knows, and an agent can act across all of it. That's the moment the platform either delivers or reveals itself as marketing. Done.ai has set the clock ticking on its own claim by shipping the first piece publicly.

For Nordic enterprise software, it's a refreshingly concrete bet. Not a frontier model, not a moonshot, just a hard, practical attempt to rebuild the boring business tools millions of SMEs use every day around what AI can now actually do. The vision is big. The proof, smartly, is being shipped one app at a time. Watch the second one closely.