Ramp launches Applied AI Solutions for finance teams

Ramp has launched Applied AI Solutions for finance teams, a new unit that will place engineers inside customer organisations to build bespoke AI systems on the Ramp platform.

The move comes as spending on AI tools rises sharply, even as many finance leaders still struggle to show a clear return. Across Ramp's customer base of more than 70,000 businesses, AI token spend has risen 13-fold since January 2025. Ramp also cited research showing that 87% of Chief Financial Officers view AI as critical, while 21% say it has produced measurable results.

Applied AI Solutions is aimed at large enterprises that treat AI as a material operating cost rather than an experimental line item. It focuses on linking fragmented financial data and business rules so that software agents can perform work that would otherwise require manual analysis or senior finance oversight.

Ramp's engineers will work directly with finance teams to map operations, systems, pain points and targets before building workflows tied to business metrics. Those workflows are designed to read from and write into a company's existing systems, including enterprise resource planning software, data warehouses, cloud storage and paper-based processes.

Internal model

Ramp presented the new service as an outgrowth of tools developed for its own finance department. It said it built what it calls a Finance Intelligence layer. This semantic structure translates general ledger accounts, internal policies and reporting logic into the operational terms used by its finance leaders.

According to the company, the internal system supports agents used for capital planning, variance analysis, board reporting and the financial close process. Those tasks had previously relied on senior judgement and manual reconciliation across several systems.

Ramp said the same structural problems recur in customer finance departments, where critical operating knowledge is often spread across an inherited chart of accounts, outdated approval thresholds, outdated vendor coding rules, and informal practices held by a small number of staff. In its view, the challenge is not limited to cleaning data but also involves capturing how a business actually works.

Implementation focus

Ramp's description of the service reflects a wider shift in the AI market towards implementation work rather than model development alone. It argued that, for many businesses, the main constraint has been the effort required to make data and business context understandable to software agents.

That has led suppliers to invest more heavily in templates, connectors and partner ecosystems, particularly in finance, where workflows are closely tied to policy, reporting structures and compliance processes. Ramp is positioning its new unit around that implementation layer, with an emphasis on building systems around each client's operating model rather than offering a standard software package.

Ramp processes more than USD $200 billion annually across its customer base. It said that transaction volume has given it insight into where finance operations tend to stall, particularly around month-end tasks and workflow bottlenecks.

Model choice

The new service will also be model-agnostic. Rather than tying customers to a single AI provider, Ramp said it benchmarks models for finance tasks and routes each production workflow to the model that performs best in terms of cost and results.

That approach is intended to address a common concern among large businesses: a system built around a single model vendor may quickly become outdated or expensive as the market changes. Ramp said it wants initial deployments to go live within weeks, starting with one workflow before expanding to other parts of the finance function.

Customers will be able to choose how much control they retain after deployment. Some may want a new AI-based interface, while others may prefer agents to work inside software their staff already use. Some organisations will take ownership after handover, while others may keep Ramp closely involved as the system expands.

Describing the challenge the unit is designed to address, Ramp said critical context in finance organisations is often scattered across systems and people rather than stored in one place. "The bottleneck was never the model but the painstaking upfront work that has to happen to make data and business context legible to agents," it said.