Everything you need to know

Vidra is an artificial intelligence company founded in 2026 specialising in the construction and industrial sectors. We build prediction models, custom AI platforms and automated analysis that run in real production. Our active projects include the Price Estimator (construction cost estimation), Area Hub (a department-based AI platform), the Tender Radar (automatic opportunity detection), the invoice and delivery note assistant, and Simply, our business management platform.

We work in three service areas (Analysis, Machine Learning and Deep Learning, and Generative AI and agents) and use them to build four types of systems:

Predictive models: cost or demand estimation and forecasting from your data, with a confidence margin on every result.

Automatic analysis: AI systems that detect, analyse and score market information without manual work.

Assistants over your documentation: ask questions about your manuals and documents and get clear answers.

Custom AI platforms: department-based management applications that grow without being rebuilt.

Primarily construction and industry. Our current clients (companies with one of our systems in production or a project under way, as of November 2026) are Argenia, Area Construcción and Garmel Instalaciones, companies in construction and building services. This specialisation gives us deep familiarity with sector-specific data, processes, and challenges.

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.

Predictive ML: scikit-learn, XGBoost, CatBoost, LightGBM and PyTorch.

Agentic pipelines: Claude and Gemini, with Python, Pydantic and OpenRouter.

RAG and chatbots: pgvector, hybrid search, Claude and OpenAI.

Platforms: React, Vite, FastAPI, PostgreSQL, Supabase, Docker and Electron.

Our API backends run on FastAPI and models are deployed on dedicated servers or within the client's environment.

All projects are fixed-scope and bespoke — the cost depends on the scope, available data, and required level of integration. We don't publish standard pricing because we don't have off-the-shelf products. To receive a proposal, use our contact form.

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.

You choose: on your servers, so your data never leaves your environment; on Vidra's servers inside the European Union, with access restricted to your team; or either of them combined with external AI APIs (OpenAI, Anthropic, Google or OpenRouter), in which case we tell you which provider is used and what data it receives. If your data cannot leave your environment, we use models that run locally.

When the system goes into production, a feedback stage begins: your team uses it day to day and, if something fails or does not fit, we fix it. After that you can hire maintenance: incident handling, updates and, for models, retraining with new data.

Both of us. We agree the metrics with you before starting, measure the system on data it has not seen and tell you its limits and when not to trust it. Your team provides the data and real cases, tests the system in its daily work and decides whether it goes into production.

Yes: we connect to what you already use (ERP, spreadsheets, email, document management or databases) through their APIs, exports or connectors. If a tool cannot be connected, we find out in the analysis phase and propose an alternative. Every project is fixed-scope, with scope, timeline and price agreed in the proposal. It can start with a pilot, and maintenance is contracted separately.

We work in four phases:

Identify: we interview your team, map your processes and calculate their real cost before you invest.

Build: we develop the solution with your data and integrate it into your current systems, starting with a pilot with measurable goals.

Adopt: we train your team to use, supervise and improve it without depending on us.

Evolve: we review the results, adjust what is needed and spot new opportunities.

We work iteratively and in direct contact with your team.

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.

Email us at contacto@vidra-ia.com, call +34 955 023 256 or use our contact form. We're also on LinkedIn at linkedin.com/company/vidra-ia. We reply within 24 hours on business days.