Finance ops · Production

How Ramp built a policy agent that handles more than 65% of expense approvals autonomously

The problem

Expense approvals were a tedious manual responsibility for managers and finance teams, and applying LLMs to a finance product required careful design to avoid losing user trust through low-quality or unexpected outputs.

Workflow diagram · grounded in source
1
Expense submitted for approval
trigger
“approving expenses. This has traditionally been a manager's responsibility but we're now applying a "policy agent" that finance teams can trust to match or exceed human judgment”
2
Policy agent evaluates with citations
ai_action
“the policy agent links directly to sections of the user's expense policy that its reasoning references”
3
Agent categorizes uncertainty
validation
“Instead, we use predefined categories which are not only easier to explain but also more accurate. This forces the model to bucket uncertainty into actionable states: users don't need a confidence score, they need to know what action to …”
4
Workflow builder gates agent action
routing
“That same workflow builder defines exactly where and when agents can act. The autonomy slider works both ways. Users can greenlight agents AND set hard stops. Not everything needs to be an LLM decision. We layer deterministic rules on to…”
5
Human escalation for uncertain cases
human_review
“When the agent isn't sure, we fall back to the pre-agent escalation process users are accustomed to”
6
Autonomous approval output
output
“we've seen more than 65% of approvals be fully handled by the agent”
7
Feedback loop refines policy
feedback_loop
“This feedback loop not only reduces the amount of work humans need to do over time, but also improves policy accuracy”
Reported outcome

Since enabling the policy agent at Ramp, more than 65% of expense approvals are fully handled by the agent, and a feedback loop continuously reduces human workload while improving policy accuracy.

Reported metrics
Expense approvals handled by agentmore than 65%
Reported stack
workflow builder
Source
https://engineering.ramp.com/post/how-to-build-agents-users-can-trust
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Since enabling the policy agent at Ramp, more than 65% of expense approvals are fully handled by the agent, and a feedback loop continuously reduces human workload while improving policy accuracy.

What tools did this team use?

workflow builder.

What results were reported?

Expense approvals handled by agent: more than 65% (source-reported, not independently verified).

How is this finance ops AI workflow structured?

Expense submitted for approval → Policy agent evaluates with citations → Agent categorizes uncertainty → Workflow builder gates agent action → Human escalation for uncertain cases → Autonomous approval output → Feedback loop refines policy.

Related finance ops cases