Engineering-led, not sales-led
Every proposal is written and delivered by the engineer who will build it. No pre-sales team promising a scope the delivery team then has to renegotiate.
About
AI architecture and automation for regulated industries.
We design and build the AI layer inside organizations that cannot send their data out: banks, asset managers, insurers, fintechs, and construction and industrial groups. The data foundation underneath, the models and retrieval on top, and the automation wired into the workflow your team runs every day.
Rexto works on one problem: getting AI past the demo stage and into a system a finance team will actually rely on. That means point-in-time correct data, model outputs that carry their own evidence, and integration into the tools compliance, risk and operations already use. Not a parallel dashboard nobody opens.
We started the firm because most AI vendors ship a working prototype and hand it off. In a regulated environment that handoff is where projects die: nobody owns the data pipeline, nobody owns the monitoring, and the model quietly drifts until an auditor asks a question nobody can answer. We stay through that part of the work, long after the demo.
The team is deliberately small: senior engineers and architects who have built trading, lending and compliance systems before, working directly with your team rather than through an account layer. Every engagement, whether a two-day architecture review or a year-long delivery, is staffed by the people who will be in the room when something breaks.
A track record before the pitch.
Valentin Badonfai
Founder, based in Dubai
Valentin founded Rexto after close to a decade in AI and data engineering, most of it spent inside organizations that could not send their data outside their own walls.
His work spans cloud infrastructure and cloud-native applications, big data platforms, and AI and machine learning including generative AI, plus the engineering leadership to take a system from proof of concept through to production.
View LinkedIn profileBefore founding Rexto, he built and led:
- 01
Hybrid AI platform, Deutsche Bank
Served a cloud-first strategy while keeping sensitive data on-premises, the on-prem and private-cloud pattern Rexto now builds around.
- 02
AuditGPT
An AI assistant for internal auditors working through complex and constantly changing regulation.
- 03
PLAIN
A government-wide platform for unified data sharing and analytics across German public administration.
- 04
Toll Collect
Real-time processing of vehicle data for one of Europe’s most complex tolling systems.
- 05
Turbine-blade production and predictive maintenance
Machine learning on industrial IoT data, from production quality to predictive maintenance, the ground Rexto’s industrial work grew out of.
Values
What guides the work.
Evidence over confidence
A model that's right 95% of the time is still wrong one time in twenty. We build evaluation and traceability in from day one, so every output can be checked against its sources.
Production is the deliverable
A working pilot on a laptop is a milestone, not the goal. We measure engagements by what's still running in your environment after we leave, not what demoed well in week three.
Data residency by default
Financial data usually can't leave the client's environment. On-prem and VPC deployment are the default architecture, not a premium add-on you negotiate for.
In numbers
The shape of the firm.
2024
Founded
3
Delivery languages
EU + ME
Delivery footprint
100%
Forward-deployed engagements