5 min read ·
News & UpdatesWe raised $3.5M because recon can’t wait for month-end
Today we’re announcing End Close’s $3.5M seed round. Foundation Capital is our largest institutional investor, joined by Y Combinator, Liquid 2 Ventures, Sunflower Capital and Angel Invest, plus experienced angel investors.
We’re going to use the money to grow our engineering, go-to-market and implementation teams, so we can build faster and get End Close in front of more of the companies that need it. The capital, and the investors providing it, also give us increased credibility with our customers who are entrusting us with their financial data.
If you’ve read our previous post, you know why we started End Close. To mark today’s milestone we want to share why the problem is getting harder, where we are now and what comes next.
Money movement is the product now
More companies now move money as part of their product: card programs, neobanks, payroll providers, marketplaces, remittance apps and payment processors. For all of them, the core promise is that money arrives in the right place, at the right time.
Real-time rails move that money around the clock, and customers expect an answer about it immediately. If someone’s remittance payment to their mother doesn’t land, they aren’t going to wait for your month-end close to find out why. They get on to support right away.
Most reconciliation tools weren’t built for that. They were built for the traditional month-end close: match transactions in batches, then hand the finance team a list of exceptions to clear manually. Breaks surface weeks after the payment should have completed. But when moving money is your product, recon is part of the product, and it can’t run on a monthly cycle.
The hard work is in the 1%
Sean spent six years at Modern Treasury, first on payments engineering and then building and running the reconciliation organization. By the end of his time there, Modern Treasury was reconciling $1 trillion a year and every morning the team still faced a queue with at least 100 new exception tickets.
The big learning from that experience was that matching transactions is the easy part. Almost all of the work is in the 1% of transactions that don’t match. A payout that bundles hundreds of payments, minus fees. A return that shows up weeks later. A customer who typed the wrong reference number. Each one means someone pulling data from three or four systems to figure out what happened to a single payment.
That’s also why David’s experience leading the error tracking product at PostHog is such a good fit for End Close. Error tracking in software is a close cousin of recon: high-volume data, things breaking in unpredictable ways, and an urgency to find out what’s going wrong.
End Close is built to deal with the 1% that generates 99% of the work. It reconciles continuously, transaction by transaction, across the banks, processors, card networks, ERPs and data warehouses you already use. The matching engine is deterministic and rules-based, so every match can be explained and audited. AI agents work on top of it, investigating exceptions, resolving the routine ones automatically and passing the rest to your team with a suggested fix.
We’re deliberate about that split. Matching millions of transactions is a job for rules, not model reasoning, and your auditors need a trail they can follow. AI earns its place in the investigative work that used to need a person, a spreadsheet and a lot of tribal knowledge. We also use it to manage the rules that you have put in place around recon, something that becomes an increasingly thorny problem as people roll on and off your recon team. Agents can do that work around the clock, and every action they take is logged.
Where we are today
End Close reconciles 99% of transactions automatically from day one. As we tune the rules to each customer’s data, that rises to 99.999% within about six weeks. Customers go live in weeks, compared to in-house builds and legacy implementations which typically take six to nine months. We’re also SOC 2 Type II compliant, verified by an independent audit.
Sunbound, which runs payments and revenue operations for senior living operators, was one of our first customers. Their problem will be familiar to anyone who receives batched payouts: a lump sum lands in the bank and someone has to work out which transactions it covers.
“As our payment volume grew, reconciling large batched payments against individual transactions became a real manual lift for our finance team and our customers. We partnered with End Close to automate those workflows instead of building them in-house, eliminating the reconciliation burden from our finance team and delivering a positive outcome for our customers.”
Thank you
Thank you to Steve Vassallo and the team at Foundation Capital, and to Y Combinator, Liquid 2 Ventures, Sunflower Capital, Angel Invest, and all our angels for backing us.
Most of all, thank you to our first customers, who trusted a very young company with their payments data, and to every finance and engineering team that has walked us through its messiest exceptions.
What’s next
We’re hiring for engineering and enterprise sales roles now.
In the coming months we’ll also release two things that should be of interest to anyone working in payments:
- RecBench, a benchmark for how well AI models handle reconciliation tasks. There’s a lot of debate about what AI can and can’t do in recon. We’d rather measure it.
- A standalone version of our agent that any team can use to start automating simple reconciliation work.
Our vision for End Close remains the same. Engineers would never run a large system without observability. We want every company that moves money to have the same view of its payments, knowing at any moment where its money is and where it isn’t. We call that financial observability.
If your team starts its day working through exceptions, we’d like to talk. Send us your historical unreconciled data and we’ll show you how much of it End Close can reconcile automatically.