Enterprise finance is being rewired from the inside. As AI tooling matures, the routine, manual work that once defined the back office — matching statements, chasing supplier queries, catching errors before payment — is increasingly being handed to software that can act on its own.

Nowhere is that shift sharper than in accounts payable, where a small percentage of misdirected spend translates into very large sums. The company at the centre of this piece has built its entire platform around the idea that autonomous agents, embedded in the payment flow, can close that gap.

An industry racing toward automation

The numbers behind accounts payable automation point in one direction: rapid, sustained growth as finance teams lean harder on AI to run their operations.

The AP automation market sits at roughly US$6.44bn today and is forecast to climb to about US$15.38bn by 2035, a compound annual growth rate of 9.1%. That expansion mirrors a broader shift across finance functions, where an estimated 76% of organisations now use AI in some form, and agentic systems capable of running multi-step processes without a human in the loop are moving from novelty to norm.

Adoption is already well underway: research suggests around 36% of enterprises have embedded or are rolling out AI agents in finance and accounting, with midsize firms reporting reductions of 40% to 60% in the time spent on manual reconciliation. One estimate from a major advisory firm puts the potential prize at US$3tn in corporate productivity gains and an average EBITDA lift of 5.4% across finance operations.

What the platform actually does

At its core, the product is an AI-powered control centre for accounts payable that sits alongside the systems finance teams already run.

It connects to existing ERP, procure-to-pay and email infrastructure, then automates the auditing, reconciliation and fraud-prevention work that has traditionally chewed up staff hours. Five modules — covering transactions, statements, a helpdesk, vendors and reporting — scan more than 500 data points on every invoice, flagging duplicate payments, posting errors, overpayments and suspicious activity before funds ever leave the business.

The founding team frames the timing deliberately. The chief executive, who co-founded the company alongside a former auditor now serving as chief product officer, describes financial technology as arriving in successive waves — and views generative AI as the biggest opportunity yet to displace manual labour in back-office finance.

A workforce of autonomous agents

The platform doesn't run on a single model — it fields a fleet of more than 20 autonomous agents, each handling a distinct slice of the AP workflow.

Take reconciliation. Matching a supplier statement against ERP records by hand can take 10 to 15 minutes per document; the statement agent automates as much as 90% of those matches. A separate helpdesk module reads incoming supplier emails, works out what's being asked, searches internal systems for the answer and drafts a reply — all without a person touching the ticket. That capability, built on cloud foundation models including well-known large language and multimodal systems, is what the company argues separates it from legacy ticketing tools that still demand manual handling of complex vendor exchanges.

The founders are candid that this only works because the company was AI-native from the start. Being built around agentic workflows, rather than retrofitting them onto older rules-based automation, is what lets the team keep shipping new products and features aimed at specific customer problems.

The whole point is straightforward — cut costs, strip out inefficiency, and lower the risk of money quietly slipping out the door. — Xelix leadership, on the platform's core promise

The backers and the names on the client list

Momentum is easiest to read through who's writing the cheques and who's running the software.

The US$160m Series B, closed in July 2025, was led by a global venture and growth-equity firm managing more than US$90bn in assets — an investment it frames as a direct bet on agentic AI reshaping enterprise finance. On the infrastructure side, a long-running relationship with a major cloud provider, dating back to 2019, supplies both the compute and the foundation models that power the agents, with the provider's solutions architects advising through development cycles.

The customer roster reads like a list of household enterprises. A global food and beverage manufacturer uses the platform to detect duplicate invoices and reconcile statements across a sprawling supply chain; two pharmaceutical giants deploy it for supplier statement reconciliation, vendor data cleansing and audit-ready compliance trails. In total, more than 200 global organisations run invoices through the system.

Why leakage is the real prize

Strip away the market forecasts and the client logos, and the business case comes down to a single, stubborn problem: money that shouldn't leave the building, but does.

Payment leakage — income never collected or funds lost through operational errors, system gaps and broken processes — remains a persistent drain. Industry studies suggest duplicate invoices, overpayments and posting errors can amount to as much as 1% of total company spend. For a large enterprise, that is not a rounding error; it is a recoverable line on the balance sheet. By positioning autonomous agents at the point where invoices are checked and payments are approved, the platform aims to intercept that loss before it happens, turning back-office diligence into measurable savings.