Custom Software
Software, AI agents and integrations for your operation. The code and data stay yours.
See the service
We build the custom software and the AI that make it possible. The code and the data are yours.








































Services
Software, AI agents and integrations for your operation. The code and data stay yours.
See the service →We prepare your data, infrastructure and team to adopt AI securely, with a clear roadmap.
See the service →Workshops built on your team’s real cases. Assessment, hands-on practice and support to apply what they learn.
See the service →Engineers dedicated to your operation. Selection, senior technical leadership and team continuity included.
See the service →Why now
Five years ago this was not an option. Today it is.
Building software stopped being a multi-year, six-figure project. That changes the question worth asking: it is no longer “which tool do I buy?” but “why am I still renting something that never quite fit my operation?”
Reason 1
AI-assisted development cut the hours needed to build production-grade software. What used to justify paying indefinite licence fees can now be built for your operation, often for less than a year of subscription costs.
Reason 2
When the system is yours, the data it produces is too: it lives on your own infrastructure, can be cross-referenced across teams, and stays available for whatever comes next. With rented software, every vendor keeps a piece of your operation.
Reason 3
The code sits in your repository and the documentation lets another team continue it. If we ever stop working together, nothing stops: there is no licence to renew and no vendor to depend on.
Reason 4
Under IFRS and US GAAP, development costs for custom software can be capitalized as an intangible asset; a SaaS subscription, by contrast, is always a period expense. That can shift part of your operating cost below EBITDA — worth reviewing with your auditor.
Reason 5
A standard system imposes its own process and forces the work to adapt to what the tool allows. A system of your own starts from the actual process: the operation defines the software, instead of the software redefining the operation.
Reason 6
A SaaS vendor has to validate that each new model works for its entire customer base before enabling it, and that takes months. On software you own, moving from one model to another is a contained configuration decision, made when it suits your operation.
Reason 7
We build on widely adopted languages and frameworks, with technical and functional documentation. Maintenance is not tied to profiles certified in one specific platform, nor to the rates of a single vendor.
Method
Step 1
We listen to the operation before proposing anything.
Step 2
We build in short cycles, with deliveries you can see and use.
Step 3
We stay while the operation changes, or hand maintenance over to your team.
Team
There are four of us. The same people who present the project are the ones who build it.

Cofounder & CEO
Industrial Engineer (U. Andes). Before founding Veltrum he worked at Milimetrix Consulting, Wolverine World Wide (Merrell) and Nestlé, building a deep view of the processes and operating challenges of large organisations. Specialised in Machine Learning, LLMs and SAP.

Cofounder & CTO
Computer Engineer (U. de Chile) with experience in machine learning, cloud and scalable systems architecture. Previously a consultant on the digital transformation of the airline LATAM.

Business Developer
Lawyer (UANDES) with a diploma in Artificial Intelligence (Oxford). Combines a solid legal grounding with AI specialisation, driving business growth and the adoption of new solutions.

Forward Deployed Engineer
Computer Engineer (U. Católica). Implements, adapts and optimises technical solutions directly in the client environment, making sure the integration lands and the systems perform.
FAQ

A 30-minute conversation is usually enough to tell whether there is anything worth building.
Book a callYou talk directly to the partners at Veltrum, not to an account executive.