Qonto's agents make bank transfers 2X faster and payroll 5X quicker, with the customer approving each
European business bank Qonto built Claude agents that prepare bulk transfers, draft invoices and answer questions about transactions. The customer confirms any action with financial impact before it happens.
- Organization
- Qonto
- Industry
- Banking
- When
- Nov 2025
- Evidence
- Vendor case study
- Agents and tools
- ClaudeAmazon Bedrock
Reported outcome
Qonto reports bank transfers 2X faster, invoice creation 3X faster and payroll 5X quicker. Customers confirm every action before it runs.
- 2X fasterBank transfers
- 3X less timeClient invoice creation
- 5X quickerPayroll
- up to €100,000Largest delegated transfers
Caveats
Published by the company that sells the agent. Treat numbers as marketing until confirmed elsewhere.
- Published by Anthropic, which sells the model, using Qonto's own numbers.
- The page gives no method or error rate behind the 2X, 3X and 5X figures.
- These compare time on a task. The page does not show the share of Qonto customers who use the agents.
- Banking reconciliation is on the roadmap, not released.
What the agent does
Qonto serves more than 600,000 businesses in eight European markets. An operator agent routes requests to task-specific agents, which prepare bulk payroll transfers and draft invoices from a client file. An analyst agent answers questions about transactions.
Complex steps run on Claude Opus and Sonnet, and simpler steps go to smaller models. On the most sensitive uses, deterministic checks sit beside the model.
What happened
The project started in November 2025, and the first agent shipped six weeks later to several thousand beta customers. Qonto reports faster transfers, invoices and payroll. It says customers have delegated transfers as large as €100,000, with the final approval always theirs.
What to copy
- Let the agent prepare and let the person approve anything that moves money.
- Show only what the person needs on the review screen.
- Put deterministic checks beside the model on sensitive steps.
Starter prompt
Written by us, not by Qonto or Anthropic. Never let a model send money without a person's confirmation and your own hard limits.
Payroll file: {rows}
Saved beneficiaries: {beneficiaries}
Prepare a draft transfer list: name, IBAN, amount, reference.
Match each row to a saved beneficiary. Flag any row with no exact match.
Show the total. Stop and ask the user to confirm. Do not send anything.
Sources
- Qonto customer story — Anthropic, 2026
Added Nov 9, 2026 · Sources last checked Nov 16, 2026.
Get new use cases weekly
One short email. Sources included.
The newsletter is paused. Past issues stay in the archive.
Related use cases
Lyft cut average support resolution time by 87% with a Claude-based assistant
Average resolution time fell by more than 87%, with thousands of requests resolved per day (company claim).
TRM Labs' security agent handled about 10,000 interactions in 30 days and closed 17% of alerts alone
In the past 30 days: about 10,000 interactions, 45% of helpdesk requests resolved with no ticket, and about 238 security alerts (17%) closed without paging anyone.
EvenUp reports personal injury drafts in about 30 minutes, down from 8 to 15 hours of skilled work
Drafting that took 8 to 15 hours of skilled work per case now returns in about 30 minutes, per EvenUp. Discovery responses went from 5 hours to 30 minutes.