Multiverse Computing and Muxunav team up on small AI models for payments and finance
Venture Map 2 firms2 outlets ShowHide
In this story 2
Coverage 2
- Forbes EspañaOriginal on Forbes España (opens in a new tab) 1 Oct
- La Ecuación DigitalOriginal on La Ecuación Digital (opens in a new tab) 2 Oct
The story so far, in brief; the legwork, and the full article, stay with the original outlets.
Multiverse Computing and Muxunav are building a new family of specialised AI models for regulated finance and unattended retail. The first place the models will run is Muxunav's payment terminals, the machines that take money with nobody standing next to them: vending machines, laundrettes, petrol stations, electric car chargers and the gates of sports grounds. Later the work is meant to reach Muxupay, the group's payments company, which is still waiting for its authorisation.
The first phase has three strands. A voice agent will guide buyers, explain products, sort out problems, check ages where the sale requires it and answer questions about allergens or charging prices. Within unattended payments it will be Muxunav's alone. If someone tampers with or breaks a machine, the system can block sales, raise alarms, log what happened and send an alert, following each operator's own rules. A second strand handles identity checks, reading documents, matching faces and screening against sanctions and politically exposed persons lists, with a human reviewing cases when appropriate. A third looks at transactions together with device, location and behaviour signals, and flags or blocks suspicious payments in real time by the client's criteria.
The models are designed to run inside the client's own infrastructure rather than as large general-purpose models rented from a third party, and will be trained on synthetic and anonymised operational data. Muxunav gets production-ready models without building an AI team of its own; Multiverse, which says it works with more than 100 organisations, gets to see how small models cope with little computing power and patchy connections before moving on to banks. "If it fits in a machine on the street, it fits in any bank," said Enrique Lizaso, Multiverse's co-founder and chief executive. The two companies expect the identity verification market to grow from about $16.5 billion in 2026 to nearly $38 billion in 2034, put worldwide card fraud losses at $33.4 billion in 2024, and estimate that the fraud detection and prevention market grows by roughly 20% a year.

