Each project starts with a concrete problem. These are the systems we have built and deployed for our clients.
Watch the video again
Simply
Constructora Meridiano
Business management platform
A platform to run a company end to end: the sales cycle (contacts, companies, offers, quotes and agreements), operations (projects by phase, tickets with SLAs and time tracking) and team organisation (calendar, meetings and events). It includes a portal where each client follows their own incidents and a real-time chat. It is multi-company: each one works in its own space, with its own users, permissions and isolated data, and adapts it with its own fields and views.
In production · Vidra's own product
In production since October 2026, with Vidra's real data. It replaces the previous internal tools and installs as an app on mobile and desktop. You can try it in the demo above.
Argenia
ML Budget Estimator
Argenia Hub · Estimator
Extensible AI hub for the construction sector. The active module estimates work item costs using machine learning models trained on real historical budget data.
Predictions with confidence intervals over Excel work item files. Available as both web and desktop app.
Area Construcción
AI Platform for Construction
Area Hub
Modular AI hub organised by departments (Studies, Procurement, Production). Designed to grow: adding a new feature means editing a configuration registry.
Finished and in production: IP-restricted web application and desktop app with silent auto-update.
Area Construcción
Tender Radar
Automated weekly review that detects, analyses, and scores private construction opportunities before they go to public tender. Uses AI to evaluate each market signal, integrated with Area Hub.
Finished system: a weekly opportunity newsletter with 0–100 score, certainty level, and inline commercial feedback.
Argenia
Price estimator
Machine learning model that estimates construction work item costs from Argenia's real historical budget data.
In use by the Argenia team: its estimates replace manual calculation as the first price reference. We keep fine-tuning the model with their feedback.
Area Construcción
Documentation assistant
Chatbot that answers questions from Area Construcción's own documentation, manuals and knowledge base. For private use by their team; it uses the OpenAI API, which does not use the data to train its models.
Finished system: answers questions about the company's internal documentation using hybrid search.
Multi-company
Invoice assistant
A system that replaces manual supplier invoice registration with an AI agent: it extracts tax ID, amount, invoice number and work-item code from every PDF, stamps the document, and leaves it ready for accounting, with a manual review queue for low-confidence cases.
In production for several companies of the same group, replacing a fully manual workflow of email, physical stamps and ERP entry.
Yes. We can start with a scoped pilot: one specific use case, on your real data, with measurable goals agreed before we begin and a duration of a few weeks. At the end we review the results against those goals together. If they are met, the pilot moves to production, integrated into your tools; if not, we explain why and what it would take. What we don't build are demos with no intention of reaching production.
It depends on scope. A predictive ML model with data already available can reach production in 6–10 weeks (exploration, training, validation, deployment). A full modular platform takes several months. We provide a detailed timeline breakdown as part of the proposal.
We are 8 people with backgrounds in AI engineering, development, design, and business. The technical team includes specialists in predictive models, automatic analysis, assistants over documentation and custom platforms.