{"schemaVersion":"1.0","language":"en","sections":[{"id":"general","title":"General","items":[{"question":"What is Captia Technology?","answer":"Captia Technology is an industrial and technology company that solves one systemic problem: operational fragmentation. We integrate machines, sensors, energy, data, ERP, processes and operations as one system rather than as isolated services."},{"question":"How is Captia Technology different from Captia AI?","answer":"Captia Technology is the company (captiatechnology.com domain). Captia AI is one of its five business units, responsible for interpreting data and automating decisions with AI. The five units are Consulting, Connect, AI, Energy and Service."},{"question":"What are the five Captia business units?","answer":"Consulting diagnoses and designs the roadmap. Connect bridges OT/IT systems and handles edge ingestion. AI interprets data and automates. Energy integrates the energy layer. Service digitalises the business layer with Odoo, apps and automations."},{"question":"What problem does Captia Technology solve?","answer":"Operational fragmentation: machines, sensors, energy, ERP and processes running disconnected from each other. Captia runs them as one integrated system, so intelligence is actionable in the real environment, not just in a slide deck."},{"question":"Which industries does Captia work with?","answer":"Captia works with industrial environments, service businesses and mid-market manufacturers across Spain and the EU. We adapt the iconography, the diagnosis and the roadmap to each industry, never the underlying architecture."},{"question":"Which technologies and programmes does Captia work with?","answer":"Captia works with established industrial technologies (Odoo, Siemens, Legrand, Alverlamp) and takes part in Spanish government innovation programmes. Those are the platforms its projects are built on; the engineering is its own."}],"description":"Captia Technology, its five units and the systemic problem it solves."},{"id":"consulting","title":"Consulting","items":[{"question":"What's the difference between diagnosis and operational audit?","answer":"A diagnosis captures the current state with zero impact on operations. An audit then validates that diagnosis against standards (ISO 50001 and others) and produces prioritised recommendations. Captia Technology always starts with a 2-week diagnosis."},{"question":"How long does it take to move from Level 1 to Level 3 maturity?","answer":"Typically 4-8 months. Phase 1 (data collection): 2 months. Phase 2 (visibility): 2 months. Phase 3 (diagnosis): 2-4 months. The timeline hinges on how much manual integration your legacy systems require."},{"question":"Does change require stopping production?","answer":"No. Captia Technology implements in parallel: the old system keeps running while we validate the new one over 2-4 weeks. We only switch once the data matches at 99%."},{"question":"What if we don't have an updated BOM?","answer":"That is common. We rebuild the BOM during the project. Captia Technology assigns an analyst who interviews the production team and documents the flow as it actually runs, not as it looks on paper."},{"question":"How do we measure program success?","answer":"With real operational KPIs: less unplanned downtime (15-30%), energy savings (8-15%), shorter cycle times (10-20%) and better quality (3-8% fewer rejects). Typical payback is 12-18 months."},{"question":"What's the typical cost?","answer":"Between €80k and €250k depending on complexity. The return comes from a more efficient operation, not from selling software. Many plants justify the investment on downtime reduction alone within the first six months."}],"description":"Diagnosis, operational audit and maturity roadmap."},{"id":"connect","title":"Connect","items":[{"question":"What is MQTT Sparkplug B really?","answer":"Sparkplug B is a specification that lives on top of MQTT. It defines data structure, delivery guarantee, strong typing and message compression. It is the standard the industry is now adopting to connect legacy machines with modern systems. Sparkplug B is to MQTT what HTTP is to TCP: a more robust protocol built for industrial use."},{"question":"Do I need to change my old machines to use unified namespace?","answer":"No. The machines stay as they are. An edge device (a small computer) translates whatever they already speak (Modbus, OPC-UA, analog signals) into MQTT Sparkplug B. The PLC stays a PLC. It simply speaks a common language the whole system understands."},{"question":"What's the difference between OT/IT bridge and simple API integration?","answer":"APIs are fragile on noisy industrial networks. One timeout and you lose data. MQTT Sparkplug B is designed to fail gracefully: if the network drops, the edge device buffers. When the network comes back, it resends. Nothing is lost silently. It is architecture, not technology."},{"question":"Does it work with machines from different vendors?","answer":"Yes. A unified namespace is vendor-agnostic. A Siemens PLC, an Allen-Bradley and a Beckhoff all publish to the same MQTT broker in the same language (Sparkplug B). That is exactly the point."},{"question":"What happens if the MQTT broker fails?","answer":"If the broker fails, the edge device keeps working offline. It buffers the data, and once the broker is back online it resends everything with no loss. That matters in plants where a communications outage must never mean lost data."},{"question":"Is MQTT scalable to 10,000 devices?","answer":"Yes. MQTT was designed for industrial IoT. Brokers such as HiveMQ handle millions of concurrent connections. Sparkplug B adds report-by-exception compression (only changed values are sent), so traffic stays manageable even with thousands of devices."}],"description":"OT/IT bridge, MQTT Sparkplug B and unified namespace."},{"id":"ai","title":"AI","items":[{"question":"How do you avoid overfitting in industrial models?","answer":"Overfitting happens when the model memorizes training data but fails in production. In industry this is critical: the model trains on one machine's data, but the machine ages. The fix is time-series cross-validation plus continuous retraining whenever the model detects drift."},{"question":"What's the difference between prediction and anomaly?","answer":"Prediction says \"this will happen in 3 days\". Anomaly says \"this is abnormal right now\". For predictive maintenance you need prediction. For safety/quality you need anomaly. Many plants mix the two up and end up drowning in false positives."},{"question":"How much historical data do I need to train a model?","answer":"It depends on the model type and the use case. For predictive maintenance, typically 12-24 months of clean history. Three years of data is better; with only two months, the model will be fragile. Captia Technology usually asks for \"at least one complete failure cycle\"."},{"question":"Do AI models work with dirty data?","answer":"It depends on how the data is dirty. Outliers (rare values) are handled well by tree-based models (random forest, XGBoost). Missing values (gaps) call for imputation. Sensor drift (calibration changes) requires manual recalibration. Captia Technology spends 30% of each project on data cleaning."},{"question":"How do I know if my model is good enough?","answer":"It depends on the use case. For predictive maintenance we want recall above 90% (catch at least 9 of every 10 failures) even if precision sits at 70% (some false positives). For quality we want precision above 95% (almost no false positives). The right metric depends on the cost of being wrong."},{"question":"What's the difference between ML and AI?","answer":"Machine learning is a technique: training models on data. AI is the broader concept. In industry, \"AI\" usually means \"ML plus business logic\". A model that predicts a failure but has no idea which technician to call is not complete \"industrial AI\". Automated decision-making is what earns the name."}],"description":"Models, data and automated decisions in industrial environments."},{"id":"energy","title":"Energy","items":[{"question":"Does ISO 50001 require investment in new equipment?","answer":"No. ISO 50001 is a management system, not a shopping list. Many savings cost nothing: switching off equipment that runs for no reason, fixing leaks, optimising compressors instead of replacing them. Captia Technology typically identifies 30-40% of the energy savings without touching the equipment."},{"question":"How long does it take to implement ISO 50001?","answer":"Certification takes 12-18 months, but savings start showing up in months 2-3. The organisational structure (owners, policies, measurement) is in place within 4-6 weeks. Everything after that is continuous improvement."},{"question":"What is IPMVP and why does it matter?","answer":"IPMVP is the international protocol for verifying energy savings. It defines how to measure the baseline, implement the improvement, measure again afterwards and calculate real savings adjusted for variables such as weather and production volume. Without IPMVP, savings are anecdotes. With IPMVP, they stand up to scrutiny."},{"question":"Do solar panels pay off without energy efficiency work first?","answer":"No. An inefficient plant wastes energy everywhere, and solar panels simply generate energy that gets wasted the same way. Optimise consumption first (a 10-15% saving). Then add renewables: they deliver more value feeding an efficient plant."},{"question":"How do I measure energy correctly?","answer":"You need smart meters at the critical points: per production line and per consumption type (motors, compressors, furnaces). A single site-wide meter is not enough. Each point needs its own measurement to pinpoint where energy is being lost. Captia Technology typically installs 8-15 meters in a small plant."},{"question":"Is energy saving a cost or an investment?","answer":"An investment. The management system, meters and improvements cost money to put in place. But the return is purely financial: money you simply stop spending. Payback runs 9-18 months, and everything after that is net gain, year after year."}],"description":"ISO 50001, IPMVP and operational energy efficiency."},{"id":"service","title":"Service","items":[{"question":"Is Odoo really an industrial ERP or an ecommerce system?","answer":"Odoo started life as an ecommerce platform. Its manufacturing modules (MRP II, WMS) exist, but they were designed without real plant experience. They work if you configure them properly and feed them real OT data. Captia Technology does exactly that, and it takes a clear architecture and middleware."},{"question":"What's the alternative to Odoo in industry?","answer":"SAP, Infor, Dassault. All of them more expensive, more complex and more mature in industrial functionality. Odoo is the right choice when you want solid functionality on a limited budget and are willing to invest in configuration and integration. It is not plug-and-play."},{"question":"Can I use Odoo if I still work with paper on the line?","answer":"Yes, but treat it as an incremental step. Digitise first: order capture, production start and end, changeovers. Without digital capture, Odoo is just a ghost copy of what actually happens on the shop floor. Captia Technology insists on real data capture before Odoo goes in."},{"question":"Does Odoo handle production variability?","answer":"Partially. Odoo handles standard orders well. When last-minute changes hit (a customer calls to change priorities, a machine fails and the schedule has to be redone), Odoo needs help: an external scheduler that understands real plant constraints makes the quick decisions, and Odoo is updated afterwards."},{"question":"How much does it cost to implement Odoo in a plant?","answer":"Between €80k and €180k depending on complexity. The software itself is cheap. The real cost sits in implementation, configuration and machine integration. Many plants fail because they expect Odoo to work out of the box, then spend twice as much on manual workarounds."},{"question":"Does Odoo replace my current ERP?","answer":"It can, but the data migration is complex. If your old ERP holds 10 years of history, that data does not just disappear. Captia Technology typically recommends running both in parallel for 2-4 weeks, validating that they match, then cutting over. No big bang."}],"description":"Odoo, digitalised business layer and plant integration."}],"lastUpdated":"2026-08-15"}