ScalePoint Research

Independent analysis of AI for payments and finance operations

ScalePoint Research evaluates AI models on payments and finance workflows, scoring them on the financial mistakes they make and the total token cost to run them. See the research agenda →

Enter research

Free to use. Sign up to get updates.

What the tests show so far

Every chart below comes from the live leaderboards and updates when a test does. Enter research for every result.

Reconciling card and payout income See the leaderboard → Reading card statements See the leaderboard → Rankings across the tests Score out of 100. Teal models have taken both tests. See the rankings → Open weights against closed Score out of 100 on each test, best model of each kind. Open-weight models can run on a firm's own servers. See hosting and terms → Where the tests sit Two of the 11 clusters of payments and finance processes have a model test so far. See the process map →

The research agenda

AI agents are showing up in payments and finance operations, both in vendor products and in tools firms build themselves. Either way, an AI model underneath does the work. The research asks two questions about those models.

  1. How well does each model do a given payments task, and what do its mistakes cost?
  2. What does each model cost to run at a firm's own volume, and where can it run with zero data retention?

The research judges accuracy in money terms. Each test has a leaderboard that ranks models on the money they get wrong, with the cost per task alongside, and is kept up to date as new models come out.

Each test takes one task from a map of 51 payments and finance processes across six types of firm, from merchants to EMIs.

Every model, every result and the full method are inside the research.

Enter research

Custom research and advisory

ScalePoint Research runs bespoke evaluations mapped to your specific processes, giving you the unit economics of AI at your real-world volumes and financial and technical clarity on:

  • Unit economics and performance: model fees plus the staff time needed to check, maintain and fix the model's work.
  • Workflow design and prototypes: how the workflow should be set up, whether to build, buy or combine, and a working prototype on your processes.
  • Vendor checks: tests of whether a vendor's agent workflow does what the demo says, and tests you can write into the SLA.
  • Regulatory readiness: practical, audit-defensible guidance on DORA, the EU AI Act and UK operational resilience rules.
Get in touch

Articles on AI in payments and finance operations

AI agents · Operating model

What Changes When an AI Agent Works the Queue

The roles, controls and numbers that change once an agent does the first pass on every case.Read → AI agents · Outlook

Will AI Agents Be Standard in Finance Operations by 2031?

How far agents have come in five years, what the forecasts say, and why adoption won't be even across processes or firms.Read → AI agents · Worked example

How an AI Agent Works a Reconciliation Break

One £1,240 settlement shortfall, followed step by step: who does what, and what drives the cost.Read → AI agents · Guide

AI Models, Copilots and Agents: What's the Difference

A plain guide for CFOs being pitched AI agents in finance and payments operations.Read →
All articles →

How to start

For custom research or advisory work, email Martin Koderisch to talk it through. Every piece of work is agreed in a fixed-price statement of work.

martin@scalepointpartners.com
Or connect with Martin on LinkedIn.