What happens without a person
Most of this category routes an invoice to a human who approves it. That is workflow, not automation, and the distinction determines whether your AP hours actually fall.
Directory · updated August 2026
This category has more marketing per unit of capability than any other in finance software. Nearly every product here captures an invoice and routes an approval; the differences are in what happens without a person and what gets recorded when something does.
Tell us your volume and vendor concentration. We will tell you the realistic automation band.
| Product | Best for | Typical price | Strength | Watch for | Score |
|---|---|---|---|---|---|
| RampRamp | $5–150M | Free–$15/user/mo | Cards, spend, and AP in one; genuinely free tier | Best where card spend dominates bill volume | 7.8 |
| erp.ioerp.io · that is us | $5–150M | Included | Agentic coding with a published authority model | Not standalone; part of a wider platform | 7.7 |
| TipaltiTipalti | $25M+, global | From $447/mo | Global payments, tax forms, mass payouts | Priced and scoped for higher volume | 7.6 |
| BrexBrex | $10–200M | Free–$12/user/mo | Strong for venture-backed and multi-entity | AP depth trails its card product | 7.5 |
| Bill.comBILL | $5–100M, broad | $45–$79/user/mo | Ubiquitous; every accountant knows it | Approval-centric; automation is thinner than marketed | 7.4 |
| StampliStampli | $10–200M | Quoted | Collaboration on the invoice itself | Pricing not published; quotes vary widely | 7.3 |
| AirbasePaylocity | $10–150M | Quoted | Spend management breadth | Direction less clear post-acquisition | 7.1 |
| AvidXchangeAvidXchange | $20M+, mid-market | Quoted | Deep in construction and real estate | Implementation is a project, not a signup | 6.9 |
We are scored on the same criteria as everyone else and we are not first. Ramp scores higher on this table because its free tier and card integration are genuinely strong for most mid-market companies.
What matters here
Capture accuracy is where the marketing is. Authority is where the difference is, and it is the question least often asked in a demo.
Most of this category routes an invoice to a human who approves it. That is workflow, not automation, and the distinction determines whether your AP hours actually fall.
Every product demos well on a clean PDF. The differences appear on a photographed paper invoice with a handwritten PO number, which is a meaningful share of real volume.
If a tool codes an invoice automatically, does the log show the reasoning, the policy, and the confidence — or only the resulting change? Only the first is reviewable.
Whether the tool pays, and if so what controls sit around vendor banking changes. That table is where business email compromise losses land.
Ask any vendor here: what can this do without a person clicking approve? The answers range from “nothing” to a bounded, configurable authority, and most sit closer to the first end than their marketing implies.
That is not a criticism of workflow products. Routing an invoice to the right approver with the contract attached is genuinely valuable. But it does not reduce AP hours the way buyers expect, because the hours were in the coding and the matching rather than in the routing.
Ramp scores 7.8 against our 7.7, and it deserves to. For a company whose spend is card-dominated rather than invoice-dominated, Ramp’s combination of cards, expenses, and AP in one product with a genuinely free tier is difficult to beat on value.
We score well on the authority and audit criteria and we are not a standalone AP product — you cannot buy our AP automation without the platform around it. For a company that wants exactly one thing solved, that is a real disadvantage and it is reflected in the score.
Any product in this category that executes payments should be asked one specific question: what is required to change a vendor’s bank details? The answer should include a second approver, a callback to a number already on file rather than one supplied in the request, and a hold on payment until both are recorded.
Almost every business email compromise loss ends at a row in the vendor master. It is the highest-value control in accounts payable and it is almost never on a feature comparison.
Our published benchmarks put bill coding at 91% straight-through at the median after twenty weeks, and 78% at the lower quartile. PO matching reaches 94%. Those are our numbers with a strict definition — no human interaction and not reversed within ninety days.
Ask other vendors for the equivalent figure and, critically, for their definition. A product counting an approval click as straight-through will quote ten to fifteen points higher for the same real performance.
Nothing. No paid placement, no referral fees, no affiliate arrangements, and no pre-publication review. Several products in this category run partner programmes that would pay us for referrals and we are in none of them — which is also why we can score a competitor above ourselves.
Questions
That one question separates workflow from automation faster than any feature comparison.