The Rise of AI-Native ERPs and What It Means for Your Finance Architecture

If you closed the books this month the way you did five years ago — pulling bank statements, matching transactions in spreadsheets, chasing approvals across email — you're not alone.  The infrastructure underneath, though, has started to shift faster than it has in decades.

In 2025, well over $300 million in venture capital went into a new generation of enterprise resource planning (ERP) startups designed for a world where AI sits at the core of how finance teams work. That kind of capital reshapes what's considered standard, and the decisions you make about your ERP and treasury infrastructure now will shape how your team operates for years to come.

Where ERP is heading

ERP has always been in motion. The move from on-premise to cloud redefined how companies deployed and managed their systems, and established platforms like Oracle NetSuite, SAP S/4HANA, Microsoft Dynamics 365, and Workday keep investing in cloud-native capabilities, AI features, and partner integrations. 

Alongside that evolution, a new class of ERP has appeared, pushed along by a few converging shifts. The biggest is that AI models reached a point in the past couple of years where they can handle core accounting workflows reliably on live financial data. The accounting talent squeeze is severe too, with 75% of CPAs set to retire within the decade. And companies are scaling across borders faster than ever, often outgrowing their accounting systems before the implementation finishes.

The new platforms are built for that reality. Implementations that used to take months can run in weeks, and month-end close has moved from a multi-week project to a few days. Transaction categorization, reconciliation, anomaly detection, and reporting all run on models that improve with use.

Who's building the new platforms

Several companies are working at this from different angles, with the shared bet that the general ledger is ready for reinvention.

Light positions itself as a "Smart Financial Platform" for companies scaling across borders, automating multi-currency bookkeeping, real-time reporting, and AI-powered pre-accounting. It raised $30 million in Series A led by Balderton Capital in 2025 and cites Lovable, KeyShot, and Xaver among its customers.

Rillet takes a "built by accountants" angle. Its Chief Product Officer is a former EY controller, and the product focuses on automated close management, GAAP reporting, and revenue recognition. The company has raised over $100 million in under a year from Andreessen Horowitz, ICONIQ, and Sequoia.

Campfire closed a $65 million Series B just 12 weeks after its $35 million Series A. Its bet is AI built into the accounting workflow itself, including LAM (a proprietary large accounting model) and Ember AI, a conversational assistant that lets finance teams query data and build reports in plain language.

DualEntry grew out of frustration. Its founders scaled a previous business to nine-figure revenue and went through what they describe as a bruising legacy ERP implementation. The result is a platform that promises to get companies live within 24 hours through an AI-powered migration engine.

What connects these companies is the conviction that finance teams deserve software built for how they work today, rather than how they worked in 2005.

What "AI-native" really means

The term gets used loosely, so it's worth being precise about it.

Adding a chatbot or a copilot to an existing ERP is valuable, and most established platforms are doing exactly this. But AI-native means something more fundamental. It means the data architecture, the transaction processing layer, and the workflow logic were all designed from day one for AI to operate on. Clean, granular data flows in real time, giving models the context they need to automate routine work.

The same logic applies to treasury, which is why Atlar built Atlar Intelligence and, more recently, AI agents for treasury that handle workflows like cash positioning and payment briefings on their own. These are capabilities built into a platform that runs on real-time bank and ERP data.

Why the bank account is a different problem

The new ERPs are doing real work on the ledger. But the moment you move from the ledger to the bank account, the nature of the problem changes.

Accounting runs on data the company already holds. Treasury runs on the bank, which sits outside the ERP entirely. Connecting to it means building and maintaining links across hundreds of banks and protocols. Cash positions are live and decay within hours. Payments are executed against external rails, with their own approval and security requirements. You need real-time cash visibility across dozens of accounts, multi-bank connectivity, payment execution with proper approval chains, cash flow forecasting, and liquidity management.

This is why no ERP, however AI-native, is built to own bank connectivity, payment execution, and liquidity on top of accounting. It isn't a feature they'll eventually ship. It's a different layer, and it needs its own infrastructure. The pattern in well-run finance teams reflects that: the ERP handles accounting and reporting, a dedicated treasury platform handles cash, payments, and liquidity, and clean integrations keep the two in sync so AI can operate across both.

Where Atlar fits in

Three of these AI-native ERPs already connect to Atlar.

The partnership between Light and Atlar put bank connectivity directly inside Light's accounting platform, so finance teams can run AI across the full workflow rather than across two systems that don't talk to each other. Lovable, one of Europe's fastest-growing AI companies and a customer of both, was among the first to go live.

Campfire and Atlar followed with a similar setup, embedding Atlar's bank connectivity inside Campfire so its close automation and Atlar's agents work from the same real-time bank and ledger data. Flex, an AI-native private bank that has raised more than $100 million, was among the early customers. As Brian Ehrlich, VP of Finance at Flex, put it: "Atlar gave us the enterprise banking layer to manage our multiple banking partners."

Rillet and Atlar connect on the same principle. Atlar links your global banking relationships directly to Rillet, so statement imports, payment runs, and reconciliation happen continuously without file transfers between the two systems. Your banking relationships stay yours; the integration is what makes them work inside the ledger.

A modular finance stack

The bigger pattern is that finance architecture is becoming modular: accounting in one layer designed for it, treasury in another, with clean data flowing between them and AI operating across both.

Whether your ERP is NetSuite, Dynamics 365, SAP S/4HANA, or one of the AI-native entrants, the treasury layer connects the same way. The only question is whether yours is keeping pace.

Teams at Lovable, Flex, Acne Studios, GetYourGuide, Mangopay, and Tide use Atlar alongside their ERPs to manage cash, payments, forecasting, and more. On G2, Atlar holds a 5-star rating, the highest of any treasury management system on the platform.

If you're weighing up ERPs, our partnerships team works across all of these platforms and can give you an impartial read on which fits your setup, including introductions where it helps. Get in touch or reach out at partnerships@atlar.com.