Predictive maintenance with artificial intelligence
Predict machine failures two weeks in advance
PRECOG learns how each electric-motor-driven machine normally behaves. It detects deviations early and tells your technicians what is likely to fail, where and when.
01Why predictive
Most maintenance still reacts to failures
If the air conditioning of a data centre fails, the whole server park can be at risk within minutes. A broken lift can block vital processes in a hospital. Every unexpected failure is costly, so maintenance is a strategic factor as well as a technical task.
Repair after it breaks
Equipment is repaired when it is already defective. It looks cheap at first, but a failed cooling system, heating system or production line causes high costs.
Unplanned outages and safety risks
Replace on a schedule
Inspections or replacements happen at fixed intervals. Unexpected failures become less likely, but many replacements are unnecessary.
Unnecessary costs and spare parts
Act when it is needed
Defects are predicted from real-time data and work is done only when it is truly necessary. PRECOG does this for each machine separately.
Planned work and continuous operation
02How it works
From raw telemetry to a clear maintenance message
Compact, task-specific models run in real time. AI classification starts only when a relevant anomaly has been detected, so every message rests on real measurements.
- 01
Measure
Energy, current, vibration, temperature or position, from existing sensors or PRECOG devices.
- 02
Analyse
14 probabilistic algorithms learn the normal behaviour of each machine.
- 03
Detect
Deviations from normal operation are identified without fixed threshold values.
- 04
Classify
AI compares the deviation with a database of validated faults.
- 05
Notify
Technicians get a short message on what, where and when, with possible repair suggestions.
Knowledge stays in the system
Technicians confirm or reject each alert with Yes or No, so the system keeps learning. The experience gained from one detected fault is transferred automatically to other, similar devices.
Adapts to each machine
Detailed modelling of every device is not needed. Tests per machine class are sufficient.
Real data only
PRECOG works exclusively with measured data, so AI hallucinations are excluded.
Indoors and outdoors
It works with continuously and periodically operating machines in both environments.
03Product
Sensors, processing and dashboard from one source
We design and manufacture custom sensor solutions, provide PRECOG AI as the processing unit and offer a dashboard for the results.
Custom sensor solutions
We design and build the measurement devices for your machines. They can also work together with the measurement infrastructure you already have.
- Energy, current, vibration, temperature, position
- Wi-Fi or Ethernet connection
- Easy installation
PRECOG AI
The analysis engine that turns telemetry into maintenance information. It learns each machine, detects deviations and identifies the likely fault.
- 14 probabilistic algorithms with AI classification
- In the cloud or on a local server
- Web API
PRECOG Dashboard
The real-time condition of the machines, the energy use and the predictions with notifications.
- Real-time condition and energy insight
- Predictions with notifications
- Cloud or on-premises, through a Web API
03Product, dashboard
Real-time monitoring and energy optimisation
The PRECOG Dashboard shows the real-time condition of the machines, the energy use and the predictions with notifications. It connects through a Web API, in the cloud or on a local server.
- Real-time monitoring of every machine
- Energy optimisation: comparison of similar machines and detection of anomalies
- Analysis of seasonality, which reveals the causes of varying consumption patterns
- Past and future energy consumption by period, in cost and in energy
- Customisable reports
- Predictions with notifications
03Product, sensors
A measurement device that fits existing infrastructure
PRECOG can build on existing measurements and also provides its own custom-built sensors. The device connects by Wi-Fi or Ethernet.
Validation
VidaSoft-hosted Web API and browser application for fast technical and business validation.
Local installation
Dedicated local PRECOG server with its own API endpoint and PostgreSQL database.
04Industries
Wherever an electric motor keeps things running
Pumps, fans, compressors, lifts, conveyors, robots and point machines, running continuously or periodically.
Manufacturing
Robotic arms, conveyors, presses, CNC centres, winding machines and paint lines.
Airports and logistics
Baggage handling, boarding bridges, moving walkways, doors and HVAC.
Hotels and buildings
Chillers, air handling units, pumps, lifts and escalators in hotels, hospitals, offices and shopping centres.
Railways and public transport
Turnouts and point machines, automatic doors, crossing barriers and station lifts.
Data centres and telecoms
Cooling, air handling and the mechanical systems behind IT infrastructure.
Energy, water and process
Pumps, compressors, cooling towers, firewater systems and process equipment.
05Project approach
We start with your most critical assets
Every project starts with a custom survey of your machines, and the price depends on the project. Failure detectability is validated on site before anything goes into daily operation.
Select the critical assets
- Analysis of the failure log database
- Failure impact, cost and frequency
- MTBF, MTTR and total downtime per asset
- Spare part criticality and lead time
Prove that it can be predicted
- Analysis of existing data
- Planning and measuring test data
- Site visit and documentation
- Validation of failure detectability
Run and keep learning
- Real-time monitoring and failure prediction
- AI-based fault identification and alerts
- Maintenance decision support
- Technician feedback and continuous learning
06Results
Measured results and accuracy
Results of a railway project that monitors the power consumption of 300 turnouts, and the accuracy of the algorithms on more than 1,500 assets and indicators.
Go deeper
Technology
Maturity levels, the algorithms, accuracy measures and a comparison with threshold methods.
Data evaluation
The input data, how results are read, prediction examples and frequently asked questions.
Periodic motion analysis
How PRECOG analyses the power curves of machines with periodic motion, with recipes.
Case studies
Six project stories, from an airport and a hotel to railway turnouts and telecommunication KPIs.
07Recognition
Independently recognised
Silver Medal among 83 inventions
PRECOG received the Silver Medal in Geneva in 2025.
“The operation management of the future will no longer mean repairing failures afterward, but knowing in advance when and where a problem is likely to occur.”
The inventor of PRECOG, in “The Future of Industrial Maintenance with Artificial Intelligence”
08About Vida Soft
Engineers and mathematicians behind PRECOG
Engineers and mathematicians designed an algorithmic system for predicting failures of electrically driven devices. It has since been refined on real machines and combined with modern AI. Vida Soft Ltd. was founded in London in 2020.
09Book a meeting
Talk to us about your machines
Choose a day and a time that suits you. We confirm the appointment by e-mail. We start with a conversation about your machines and then prepare a custom survey and a project-based offer.