Introduction: An Exception to “Never”
In May 2025, Sequoia Capital did something it had never done before: it invested in a general ledger software company.
Sequoia partner Julien Bek was widely quoted by tech media: “ERP is one of the largest software categories, but it has barely changed over the last decade because rebuilding a company’s financial foundation is extremely complex.”
Even more interesting was the involvement of Sequoia managing partner Roelof Botha. Botha is PayPal’s former CFO, so he understands the pain of finance software from the inside. In Rillet’s announcement, he said Rillet had rethought the general ledger, using real-time integrations and AI-driven workflow automation to help finance teams work smarter.
That $25 million Series A came only ten months after Rillet’s previous round. Less than half a year later, a16z and ICONIQ Growth co-led a $70 million Series B.
Eighteen months, $95 million, and backing from Silicon Valley’s top venture firms. How did an ERP company that only launched in 2024 earn that kind of attention?
1. Founder-Market Fit: Accountants Building for Accountants
Rillet CEO and founder Nicolas Kopp has a sentence that defines the company’s DNA: “We are a system built by accountants for accountants.”
That sounds like a polished slogan, but the product makes it feel literal.
In a Rippling interview, Kopp admitted that one of his early hiring mistakes was overvaluing “learning ability” and undervaluing domain expertise. He assumed smart people could learn accounting quickly. They could not. Accrual accounting, reconciliation logic, and GAAP rules are not things people master through raw intelligence alone.
That lesson shaped Rillet’s team: accounting technologists plus software engineers. The Org lists the company at roughly 11-50 employees and shows multiple core team members with CPA backgrounds.
The investor list reinforces the point. Former NetSuite CFO Ron Gill invested personally as an angel. A former NetSuite CFO backing a company trying to replace NetSuite is about as strong an industry endorsement as a startup can get.
2. Core Product: Not AI on Top of Old Systems, but Rebuilding from the Ledger
To understand Rillet’s product logic, start with one question: why not simply add AI to NetSuite?
The Problem: 1990s Architecture Blocks AI
Traditional ERP systems, such as NetSuite founded in 1998 and SAP founded in 1972, were designed around human operation:
- Data is scattered across modules and databases.
- Reporting depends on ETL pipelines and Excel workarounds.
- APIs are old, and real-time access is hard.
- Audit trails were not designed for executable AI agents.
Kopp’s view is that AI needs a single source of truth: real-time, clean, structured data. Legacy systems were not built for that underlying data model.
The Solution: Rebuild from the General Ledger
Rillet is attempting a clean start:
-
A redesigned general ledger data model: Instead of simply migrating old records, Rillet redesigns the chart of accounts and journal-entry structure so AI can natively understand account meaning and relationships between entries.
-
Aura AI Agents: AI agents execute accounting workflows such as accruals, reconciliations, bank reconciliation, revenue recognition, and board reporting. This is not a generic chatbot. It is a specialized agent embedded with accounting logic.
-
Aura Flows plus Close Checklist: This may be Rillet’s sharpest product decision. AI workflows are not placed in a separate chat window. They are embedded directly into the close checklist accountants already use every day.
Imagine an accountant opening the system in the morning and seeing a task: “reconcile AR aging.” One click lets the AI pull the AR aging report and the general-ledger accounts-receivable balance, compare the two, and flag differences. The accountant reviews, signs off, and completes the item without leaving the checklist.
- Continuous Close: Traditional month-end close exists because data updates are slow and reconciliation is batched. When all data posts in real time and AI reconciles continuously, month-end close becomes continuous close: the books are effectively being closed every day.
Why This Matters More Than an AI Plug-in
Because embedding determines adoption.
Most AI accounting tools ask users to open a chat window, enter instructions, and copy results back into their workflow. Rillet treats that as a parlor trick. Useful accounting AI does not wait for you to ask. It takes over work you already do and returns the result inside the place where you work.
The product philosophy is clear: AI should not create a new interaction pattern. It should make the old one disappear.
3. The Key Productization Bet: Why AI Makes Replacing Accounting Systems Possible
This question deserves its own section because it answers the deeper “why now?”
Rillet is betting on three conditions converging.
Condition 1: AI Creates a Reason Companies Must Switch
In the past, companies had little reason to replace ERP. A new system might be 10% better, but data migration, employee training, and business disruption were too expensive. Marginal improvement was not enough.
AI agents require conditions legacy ERPs cannot provide: a single source of truth, real-time data, and structured context. That means using AI becomes the reason to switch systems, rather than merely a benefit after switching.
Rillet’s comparison page puts it sharply when describing SAP: “Legacy system with AI taped on.”
Condition 2: The Mid-Market Gap
NetSuite has a strong position in $1 billion-plus enterprises, and QuickBooks dominates businesses under roughly $5 million in revenue. But companies with $10 million to $500 million in annual revenue are underserved:
- Their accounting complexity has outgrown QuickBooks: multi-entity structures, multi-currency operations, and complex revenue-recognition rules.
- They cannot justify NetSuite’s cost and implementation cycle: six to twelve months of deployment, custom scripts, and expensive consultants.
- They often live in a patchwork of QuickBooks, Maxio, Excel, and plug-ins, leaving data fragmented.
Rillet’s positioning is direct: “Go live in weeks with Rillet’s CPA-led team.”
Condition 3: AI Lowers Product-Building Cost, but That Is Not the Core Moat
AI may reduce the cost of building certain modules, but for Rillet, the real moat is not AI itself. It is accounting domain knowledge and customer migration cost.
As Kopp learned, AI is useful, but if you do not understand revenue recognition under accrual accounting, it can produce beautifully wrong answers.
4. Commercialization and Pricing
Rillet does not disclose public pricing, which is typical for an enterprise, sales-led model. Based on its positioning and competitors, a reasonable inference is:
- Annual subscriptions tiered by users, entities, and transaction volume.
- Entry pricing likely in the $15,000-$50,000 per year range, higher than QuickBooks’ roughly $1,000-$5,000 per year but far below NetSuite’s $100,000-plus annual cost.
- Advanced modules such as Aura AI Agents, advanced revenue recognition, and multi-entity consolidation may be priced separately.
In less than two years, Rillet has reached:
- Nearly 200 paying customers.
- 5x year-over-year revenue growth.
- Billions of dollars in transaction volume processed.
Customers include Windsurf, Decagon, Postscript, Kickstarter, Smartcar, Scribe, Luxury Presence, and other notable technology companies.
5. Competitive Landscape: Rillet Versus Four Camps
Rillet has comparison pages for each major competitor. That is unusual in B2B software and signals confidence in its differentiation.
| Dimension | Rillet | NetSuite | Sage Intacct | QuickBooks | SAP |
|---|---|---|---|---|---|
| Positioning | AI-native ERP | Legacy ERP | Cloud accounting | Small-business accounting | Enterprise ERP |
| Target customer | Mid-market | Mid-to-large | Mid-market | Small businesses | Large enterprises |
| AI capability | Natively embedded with Aura | Shallow integration | Limited | Limited | “Taped on” |
| Implementation | Weeks | 6-12 months | 1-3 months | Immediate | 12+ months |
| Multi-entity support | Native | Requires configuration | Requires configuration | Not supported | Native |
| Pricing | Not public, mid-high | High | Mid-high | Low | Very high |
Rillet’s core competitive narrative is that old systems require “custom scripts and workarounds,” while Rillet replaces them with “native integrations.”
6. Growth Flywheel and Distribution
Rillet’s growth flywheel has several important gears:
-
The Sequoia/a16z portfolio effect: Endorsement from top firms helps Rillet become a default vendor candidate for their portfolio companies. Customers such as Windsurf and Decagon show this effect.
-
CPA partner network: Rillet has both accounting partners and investor partners. Accounting firms recommend Rillet when helping clients choose systems. Investors help portfolio companies automate finance.
-
EY strategic alliance: In April 2026, Rillet announced an alliance with Ernst & Young. Endorsement from one of the Big Four accounting firms gives Rillet a trust credential for the enterprise market.
-
Content plus community: Close Club serves finance professionals, the blog publishes AI accounting content, and each competitor page is optimized for SEO.
-
A “no going back” experience: Once customers experience continuous close instead of month-end close, returning to the old cadence feels impossible. That is the ideal retention mechanism for an AI product.
7. Builder Lessons
1. AI-native is not a feature. It is an architecture decision.
Rillet’s biggest strategic judgment is that NetSuite cannot simply be upgraded with AI. The underlying data model must be rebuilt. That is painful, because it means a longer development cycle and higher upfront cost, but it creates the long-term moat.
Question to ask yourself: Is your AI product truly rebuilding the logic of the workflow, or is it putting an AI interface on top of the old process?
2. AI should disappear into the workflow, not become a new workflow.
Rillet embeds AI into the close checklist instead of creating a standalone AI chat window. The decision looks simple, but it has enormous impact on adoption.
First principle of AI productization: Users should not have to learn how to use AI. AI should appear where they already work.
3. Domain expertise beats general intelligence.
Kopp’s founder story returns again and again to one theme: he initially underestimated the barrier of accounting expertise. General-purpose AI does not understand accrual systems and GAAP rules well enough on its own.
Implication: If your AI product solves a professional-domain problem, your team needs people who know that domain deeply. LLM capability alone is not enough.
4. Find the experience users cannot return from.
Continuous close is powerful because once a customer experiences it, month-end close feels obsolete. That is the north star for AI products: create an experience users cannot imagine giving up.
5. Position precisely between giants.
Rillet is not trying to replace SAP, where enterprise complexity is punishing, and it is not trying to replace QuickBooks, where small businesses have little budget. It targets $10 million-$500 million mid-market companies: underserved, complex, and willing to pay.
8. What to Watch Next
- Enterprise breakthrough: Can Rillet use the EY alliance to enter $500 million-plus enterprise accounts?
- Product expansion: Will it move into financial planning, tax, internal audit, or adjacent finance modules?
- Competitive response: Can NetSuite and SAP iterate AI features fast enough, or does their architecture prevent true AI-native behavior?
- China implications: Do Yonyou and Kingdee face the same AI-native replacement window? Could a Rillet-like startup emerge in China?
Conclusion
Rillet is not just another AI funding story. It reveals a larger industry cycle: when a general-purpose technology such as AI creates requirements that conflict sharply with legacy infrastructure, a window opens to replace the entire system.
In that window, the winners are not the companies that add AI to old systems. They are the companies that rebuild systems for AI.
For product builders asking “what can AI do?”, Rillet offers a clearer route: do not ask how AI can make your product more useful. Ask what conditions AI needs to create maximum value, then build those conditions.