How 1,000+ teams automated expense reviews — and eliminated hidden risk

Reviewing expenses manually used to be the best option for teams to stay in control. Today they’re a blind spot.

Receipts pile up. Approvers apply policies inconsistently. Edge cases slip through. By month-end, controllers are left hoping nothing material was missed.

That’s why, less than six months after launch, more than 1,000 finance teams have stopped relying on manual reviews.

They’ve handed first-pass review to the Ramp Policy Agent, an always-on AI reviewer that enforces your policy consistently, across every transaction, with full organizational context.

Teams using the Agent are:

• Reclaiming 4-5 hours per week from manual reviews
• Catching 7x more out-of-policy spend

Some real-examples of policy exceptions caught include:

• Personal Amazon purchases mixed into a business order
• Flights booked outside a customer’s designated 21-day advance purchase window
• Inaccurate memos that contradict receipt data
• Hinge subscriptions on a SaaS card (yes, really!)

For controllers, the shift is becoming clear: manual reviews aren’t just inefficient; they’re risky.

Why first-pass review can’t be a human job

Manual expense review breaks down because policies aren’t enforced consistently or updated often enough.

Imagine an employee submits a flight expense and the receipt includes a seat upgrade. In a traditional workflow, what happens next depends on the reviewer:

• One reviewer approves it, assuming upgrades are in policy
• Another rejects it, unsure what rule applies in this situation
• A third approves it just to clear the queue

Same policy, different outcomes.

Over time, this inconsistency quietly erodes compliance. Out-of-policy spend slips through, employees get mixed signals, and controllers are cleaning up exceptions after the fact instead of preventing them.

The Policy Agent removes that subjectivity. It evaluates every transaction and recommends:

1. Approval
2. Rejection
3. Review

Crucially, it shows which policy rules it’s relying on, and flags where those rules are outdated or unclear.

For example, if seat upgrades aren’t explicitly addressed, the Agent escalates the expense and surfaces the gap. The Controller updates the policy:

“Seat upgrades <$200 are approved only for specific roles.”

This evolves your policy from a static PDF to a living, breathing document. Private notes like “VP level and above” guide the Agent without exposing internal nuance to employees.

From that point on, enforcement becomes automatic and consistent — and because the Agent is native to Ramp, it can apply policies with context that bolt-on tools can’t match.

For Glia, a software call center solution, that built-in context has been critical:

“We provide mobile reimbursements only for certain people and teams, and I didn’t want that in our employee-facing policy. With hidden notes, the Agent follows those nuances. It flags reimbursements that shouldn’t be paid and even catches duplicate receipts. I wrote ‘sales team’ in the policy and it knew which employees that meant.”
—Meng Murphy, VP of Accounting, Glia

Humans focus only on the exceptions

As policies get refined, most expenses stop needing human review at all.

Controllers decide:

• Which expenses the Agent can automatically approve when clearly in policy
• Who exceptions should escalate to when they’re not

The result: reviewers focus only on the 10-15% of transactions that actually require judgement.

The next time a VP books a seat upgrade in line with what’s in policy and under the approval threshold, the Policy Agent automatically approves it — no queue, no reminder, no bottleneck.

For Highspot, this shift eliminated a major pain point:

“Before the Policy Agent, we had hundreds of unapproved expenses, and it was an accounting nightmare. Now managers review a handful of transactions instead of dozens or hundreds.”
—Julia Braun, Senior Director of Enterprise Procurement, Highspot

Compliance becomes proactive, not reactive

When policy enforcement is consistent, finance teams can move from reacting to individual expenses to making strategic decisions based on trends.

Instead of asking, “Why was this approved?”, teams can ask, “Why does this keep happening?”

Admins see where violations cluster, how often reviewers disagree with the Agent, and which teams or expense types drive exceptions.

This is the new normal, and it’s now available on Ramp Plus

More than 1,000 teams have already made the shift.

At this point, manually reviewing every expense isn’t just inefficient — it’s choosing inconsistency, wasted time, and unnecessary compliance risk.

Leaving the status quo means:

• Reclaiming hours to focus on high-value work
• Saving money from policy exceptions
• Ensuring compliance with full audit trails and reports

The real question isn’t whether AI should review expenses but why anyone would still do it the old way.

As Philip Parsons, Controller at a leading enterprise customer experience platform, puts it:

“I’m excited for the agentic direction Ramp is going…having an assistant on Ramp to help myself, our staff accountants, our CFOs, and our reviewers — what a better way to use AI.”