Written by: Kamil Ponicki, Director of Talent Acquisition, Digital Colliers
Rillet just closed a $100M Series C at a $1B valuation. The product is AI-native general ledger software aimed squarely at the mid-market. If ledger tooling, which is about as unsexy as software gets, is now a unicorn category, that tells you something about where the money thinks finance ops is heading.
And if you're a CFO running NetSuite plus a stack of point tools bolted around it, the message is uncomfortable. The vendors who make those point tools are the ones about to feel the squeeze. So are the finance teams who built their process around them.
What AI-native accounting is actually eating
The interesting thing about Rillet isn't the valuation, it's the surface area of the pitch. It's not close automation as a bolt-on. It's the ledger itself, with AR, AP, revenue recognition, reconciliations, and close all sitting in one system that assumes an LLM is in the loop by default.
That's a direct threat to a whole category of tools mid-market finance teams have been stitching together:
- AR automation layers sitting on top of NetSuite or Sage Intacct
- AP and invoice capture products doing OCR plus approval routing
- Standalone reconciliation tools for bank, intercompany, and subledger tie-outs
- Close management software that's basically a checklist with Slack notifications
- Flux and variance analysis tools that pull GL data into a separate UI
Each one of those was a reasonable purchase in 2019. In 2025, the question a smart CFO is asking is whether the underlying ledger should just do that work natively. Rillet's raise says at least one tier of investors thinks the answer is yes.
The gap this exposes in mid-market finance
Here's the awkward part. Best-in-class finance teams close the books in under 5 days. The typical mid-market team runs 8 to 10 days, and most of them still get there by pulling numbers across systems in spreadsheets by hand.
That gap isn't a tooling gap in the sense of "buy another SaaS." It's a stack coherence gap. Every point tool you added to shave a day off close also added an integration to maintain, a reconciliation to run between systems, and a place where the audit trail gets fuzzy. The AI-native pitch is that a single system with the data in one place can do in one pass what your current setup does in five.
For financial services operators, this hits harder. You're already carrying DORA obligations that have been in force since 17 January 2025, which means every third-party system in your finance stack is now something you need to inventory, risk-assess, and prove operational resilience for. Eight point tools is eight DORA workstreams.
A 90-day stack review that isn't theatre
Before anyone rips out NetSuite, do the boring work. Here's a review you can actually run in a quarter.
Days 1 to 30. Map what you've got.
- List every finance tool with a contract, owner, renewal date, and annual cost.
- For each one, write down the single job it does that the core ERP doesn't.
- Time-box your close. Where do the days actually go. Which steps are humans copying data between systems.
Days 31 to 60. Pressure-test the point tools.
- Which tools are solving problems the ERP roadmap will handle within 12 months.
- Which vendors have shipped meaningful AI capability in the last two releases versus which are pasting a chatbot on the login screen.
- Which integrations break at month-end and quietly cost you a day.
Days 61 to 90. Decide, don't pilot forever.
- Kill one to three tools where the ERP or an AI-native replacement covers 80% of the job.
- Pick one workflow, close, reconciliations, or AP, to rebuild around a single system in the next two quarters.
- Set a target close time. If you're at 10 days, aim for 6. Do not aim for 5 in year one.
Where the AI story gets you in trouble
One caution. Around 95% of enterprise AI projects don't reach production or ROI. The AI-native accounting wave will have the same failure rate as every other wave.
What that means in practice is that you should not buy on the demo. The vendors who survive this cycle will be the ones whose product still works when your controller pushes back on a suggested journal entry and wants to see the source data in three clicks. That's a boring test. Run it anyway. The teams that end up looking smart in 2027 are the ones who ran boring tests in 2025.