Predict failures before they occur, detect defects with cameras, optimize production with real-time data, and keep your technicians connected with voice assistants that know every machine.
In manufacturing, downtime is enemy number one. Every minute a line is stopped is money that doesn't come back. Every defect that passes is a customer who may not return. Every technician in transit is a technician not repairing. Precise production and preventive maintenance are not options: they are competitive survival.
Voyant connects vibration, temperature, and electrical current sensors to your machines. Detects anomalies in real time, predicts failures 2-4 weeks before they occur, and alerts with enough time to schedule maintenance without stopping the line.
Industrial cameras with SomaTech AI inspect each part in real time: detect cracks, deformations, incorrect size, defective finish. They don't get tired, don't have bad days, don't vary their criteria. 100% inspection, 0% fatigue.
SomaAgent01 knows every machine, every manual, every historical failure, every procedure. A field technician asks by voice how to calibrate CNC lathe #3 and receives step-by-step instructions with references to the technical manual. Knowledge never retires.
Voyant connects vibration sensors (FFT analysis to detect unbalance, misalignment, damaged bearings), temperature sensors (overheating alerts from excessive friction), and electrical consumption sensors (detection of anomalous peaks indicating growing mechanical resistance). Data is analyzed every 5 minutes with machine learning models that learn the normal behavior of each machine. When a deviation exceeds the 95% confidence threshold, the system alerts 14-21 days in advance, identifying the likely failed component and suggesting spare parts and repair procedures. Reduces unplanned stops by 70%.
Industrial vision system with high-resolution cameras mounted at critical points on the production line. Each part is photographed from multiple angles and analyzed by convolutional neural networks specifically trained to detect: surface and subsurface cracks, dimensional deformations (±0.05mm tolerance), inconsistent surface finish, contamination or residues, and incorrect assembly. Inspection takes 0.3 seconds per part. Defect detection rate of 99.2%, with false positive rate of 0.8%. Quality data is stored traceably for audits and ISO certifications.
Voyant calculates OEE (Overall Equipment Effectiveness) in real time: Availability = actual production time / planned time; Performance = actual speed / nominal speed; Quality = good parts / total parts. The dashboard shows current OEE by machine, by line, and by plant, with alerts when it falls below configurable thresholds. Identifies the top 3 causes of efficiency loss per shift. Enables operational decisions in minutes, not weeks.
SomaAgent01 + AgentVoiceVox acts as a field technical assistant: the technician asks by voice how to repair failure F-234 on the Atlas Copco compressor and receives step-by-step instructions with references to the service manual. Can send a photo of a damaged part and the system identifies the component, searches for the equivalent spare part in the catalog, and confirms warehouse availability. Reduces diagnostic time by 60% and eliminates unnecessary trips by having remote diagnosis before departure.
Voyant analyzes real cycle times, setup times, historical failures, raw material availability, and customer orders to generate optimized production scheduling. Minimizes changeover times by grouping similar batches, considers predictive maintenance windows in scheduling, and adjusts line speed based on workload. Reduces work-in-process inventory by 30% and improves delivery punctuality by 45%.
SomaBrain ingests all technical manuals, maintenance procedures, historical failure records, training videos, and notes from experienced technicians. Builds a vector knowledge base in 8,192 dimensions where every technical query finds the most relevant information in milliseconds. When a new technician asks how to adjust line 3's PLC, the system finds the exact procedure, the last time it was done, who did it, and what problems arose. 20 years of knowledge is not lost when a technician retires.
From the sensor on the machine to the supervisor's decision, every layer is optimized for continuous production.
The factory brain: IoT sensors, real-time OEE, predictive maintenance, and production optimization.
The technical assistant: knows every machine, manual, and historical failure. Answers by voice with precise instructions.
The technical memory: every manual, procedure, and 20 years of experience, accessible in seconds.
The technical support line: remote voice diagnosis with access to machine history and spare parts.
The long-term memory: remembers optimal configurations for each product, batch, and season.
We install vibration, temperature, and current sensors on your critical machines. Configure vision cameras at key inspection points. Connect existing PLC and SCADA for production data extraction. All without stopping your lines.
We ingest all your technical manuals, procedures, failure records, and technician notes. SomaBrain builds the complete technical knowledge base of your factory, indexed and semantically searchable.
Voyant learns the normal behavior of each machine over 2-4 weeks, establishing baselines for vibration, temperature, electrical consumption, and cycle times. Vision networks are trained with samples of good and defective parts.
The system monitors 24/7, predicts failures weeks in advance, inspects every part in real time, and assists field technicians with voice and vision. You produce more, stop less, and maintain consistent quality.
SomaTech Ecosystem
13 products. One vision.
Fewer unplanned stops
AI vision defect detection rate
Improvement in delivery punctuality
Less technical diagnostic time
Less work-in-process inventory
Continuous monitoring of critical lines
Complete quality traceability by part, with photographic records and automatic inspection data. Preparation for ISO 9001 audits and customer quality certifications.
Monitoring of safety conditions: anomalous temperatures, dangerous vibrations, electrical conditions out of range. Automatic alerts before incidents.
Energy consumption and emission control. Process optimization to reduce waste and carbon footprint. Compliance with Ecuadorian environmental regulations.
Auditable code for customer audits and certifications. No vendor lock-in dependencies that compromise operational continuity.
The main motor's vibration sensor detects an anomalous pattern: the FFT spectrum shows frequencies at 58 Hz and 116 Hz, indicative of a bearing with damaged race on the opposite side. Voyant alerts 18 days before failure: SKF-6214 bearing on drive shaft, failure probability 94%. Spare available in warehouse. Maintenance scheduled for Saturday without stopping the line. On Saturday the bearing is changed in 2 hours. The line never stopped unexpectedly. $35,000 in production loss was saved.
On the plastic injection line, the vision camera detects a 0.08mm deformation on part #45,892 of Toyota's batch. The system automatically rejects the part, alerts the operator, and adjusts the mold temperature parameters. The complete batch is inspected with increased sampling. No defective part reaches the customer. Toyota never knows they almost received out-of-tolerance parts. The quality reputation remains intact.
The Ambato plant reports that hydraulic press PH-7 doesn't close correctly. Instead of sending a technician on a 3-hour trip, the supervisor calls AgentVoiceVox. The system accesses PH-7's history, identifies that the last similar failure was 8 months ago due to residue buildup in the check valve. Guides the local operator step by step through cleaning and recalibration. The machine operates in 45 minutes. Saved 6 hours of travel, $120 in fuel, and a lost production day.
Doesn't depend on a single vendor that could raise prices, discontinue products, or stop support. You control your technology and can modify it as your processes evolve.
Connects with any PLC, SCADA, ERP, or industrial sensor. Not limited to proprietary protocols. Your existing equipment investment is protected.
Each factory is different. You can train specific models for your products, adjust alert thresholds according to your tolerances, and customize dashboards for your KPIs.
Team in Ecuador that understands the local industrial reality: suppliers, spare parts, regulations, and the urgency of a stopped line.
Self-hosted
Download the code from GitHub. Deploy on your industrial server or edge device. No license costs per machine or per sensor. Ideal for plants with their own technical staff.
Download from GitHubManaged
Complete implementation with sensor installation, vision camera calibration, PLC/SCADA/ERP integration, technician training, and 24/7 support with SLA response time. Price on request.
Contact SalesBook a 2-hour technical visit. We evaluate your critical lines and present a predictive maintenance and AI quality control plan, no commitment.
