- Subject
- Microsoft 365 & SharePoint
- Filed
- Source
- Laura Varley
- Loans
- 518
End Close raises $3.5m to expand payment reconciliation platform
The Irish-founded company will grow engineering, sales and implementation teams and plans two releases focused on AI-assisted reconciliation.

End Close is putting a $3.5m seed round behind a specific growth plan: expanding the teams that build, sell and implement its payment reconciliation platform, while preparing two new products for release in the coming months. For finance and payments teams handling high transaction volumes, the company’s approach pairs rules based matching with AI agents that investigate exceptions.

Funding supports hiring and delivery
Founded in early 2026 by CEO Sean Bolton and CTO David Newell, End Close is headquartered in San Francisco and was founded by Irish entrepreneurs. Its seed round included its largest institutional investor, Foundation Capital, alongside Y Combinator, Liquid 2 Ventures, Sunflower Capital, Angel Invest and experienced angel investors. End Close was part of Y Combinator’s winter 2026 batch.

The company plans to use the capital to add staff in engineering, go to market and implementation. It is currently recruiting for engineering and enterprise sales roles. Those hires point to two needs common to a growing enterprise software business: extending the product and supporting customers as they connect it to existing payment and finance systems. End Close has not disclosed a hiring target or a timetable for filling the roles.

Implementation will matter because reconciliation depends on bringing transaction records together across multiple systems. End Close says its platform works across banks, processors, card networks, enterprise resource planners and data warehouses. For customers, the practical question will be how well those connections fit their own payment flows and data formats; the company has not specified supported vendors or integration requirements in the information available.

Rules handle matching; AI agents investigate exceptions
End Close describes its core matching engine as deterministic and rules based. It compares transactions across connected systems and records the basis for each match, with the aim of making results explainable and auditable. The company says the platform reconciles continuously, transaction by transaction, and is designed for high volume payment businesses.

The design separates routine matching from exception handling. End Close says AI agents sit on top of the rules based engine to investigate mismatches, resolve routine exceptions automatically and send the remaining cases to staff with a suggested fix. That division is intended to keep the matching decision tied to defined rules, while using AI to help teams work through cases that need investigation.

That distinction will be important for finance operations teams assessing automation. A system can reduce manual review only if staff can understand why records were paired, see which exceptions were handled automatically and review unresolved cases. End Close says its matching produces an audit trail and that exceptions can be passed to the team, but has not provided performance figures, accuracy measurements or details of how customers configure rules and review agent actions.

The company frames its target as the small share of transactions that creates most of the reconciliation workload. In practice, the value of that approach will depend on how often exceptions are correctly classified, how readily suggested fixes can be checked and whether teams can maintain oversight as transaction volumes grow. Those are the operational details payment companies will want to evaluate as the platform develops.

Two planned releases extend the product
End Close plans to release RecBench in the coming months, describing it as a benchmark for measuring how well AI models handle reconciliation tasks. It also plans a standalone version of its agent, intended to let teams automate simple reconciliation work. The company has not given release dates, described the benchmark’s test method or identified which models will be assessed.
RecBench could give teams a way to compare model performance on reconciliation work, while a standalone agent would offer a separate route to automating simpler tasks. The practical value of either release will depend on the evaluation criteria, the systems the agent can work with and the controls available to staff. Until those details are shared, teams should treat the plans as upcoming product work rather than available capabilities.
Leadership brings payments and product experience
Bolton spent six years at Modern Treasury as a tech lead in payments engineering before building and running its reconciliation division. Newell previously led the error tracking product team at PostHog, founded Unflow and worked as a product engineer at Intercom and Basecamp. That mix gives the founders experience in payment infrastructure, reconciliation and software product development.
For payment operators considering the platform, the next useful signals will be the announced releases and clearer information about integrations, exception controls and measured results. The $3.5m round is intended to fund the staff and product work behind that next stage.