Our works
We have implemented various projects using artificial intelligence, computer vision, and advanced language models.
Pallet Recognition System for AGV
Client/Industry: Leading mobile robots/AGV manufacturer
Problem: Difficulty in precise recognition of unevenly placed pallets in industrial environments.
Goal: Develop a vision system for accurate pallet detection and localization.
Solution: We implemented an advanced vision solution that enables precise recognition and localization of unevenly placed pallets in pallet nests. The integration of machine vision algorithms with autonomous forklifts increased operational efficiency and safety in industrial environments.
Problem: Difficulty in precise recognition of unevenly placed pallets in industrial environments.
Goal: Develop a vision system for accurate pallet detection and localization.
Solution: We implemented an advanced vision solution that enables precise recognition and localization of unevenly placed pallets in pallet nests. The integration of machine vision algorithms with autonomous forklifts increased operational efficiency and safety in industrial environments.
Results: 95% Fewer Warehouse Accidents: The AI-powered detection system significantly reduced safety risks and improved operational workflows. Cost Savings: The outsourcing approach saved 35% compared to building an in-house AI team.
Empathiq – Reputation Management Platform
Client/Industry: Healthcare / San Diego, California
Problem: Inability to effectively monitor and respond to customer opinions across multiple channels.
Goal: Create a real-time sentiment analysis system integrated with existing CRM infrastructure.
Solution: We developed a comprehensive platform using NLP and advanced sentiment analysis to monitor customer opinions in real-time. The system enables quick response to feedback and integrates with existing CRM systems, providing companies with key indicators regarding brand reputation.
Problem: Inability to effectively monitor and respond to customer opinions across multiple channels.
Goal: Create a real-time sentiment analysis system integrated with existing CRM infrastructure.
Solution: We developed a comprehensive platform using NLP and advanced sentiment analysis to monitor customer opinions in real-time. The system enables quick response to feedback and integrates with existing CRM systems, providing companies with key indicators regarding brand reputation.
Results: Within the first three months of implementation, response time to negative feedback decreased by 65%, while the customer satisfaction score (CSAT) increased by 22%. The system detected an average of 93% of key brand mentions in real time, enabling the customer support team to proactively address issues. CRM integration improved team efficiency by 30%, thanks to automated case tagging based on customer sentiment.
Fork Recognition System for AGV
Client/Industry: Industrial logistics
Problem: Forklift forks were often undetected by standard sensors, leading to AGV (mobile robots) collisions and safety hazards.
Goal: Develop a reliable AI-powered detection system for forklifts' forks.
Solution: We developed an AI-powered detection system using YOLOv5 and Faster R-CNN with 95% precision and real-time performance (50ms). The solution included a mobile app for live detection optimized with ONNX Runtime, and extensive image collection and annotation for training. The system was integrated with IoT infrastructure using AES-256 encryption
Problem: Forklift forks were often undetected by standard sensors, leading to AGV (mobile robots) collisions and safety hazards.
Goal: Develop a reliable AI-powered detection system for forklifts' forks.
Solution: We developed an AI-powered detection system using YOLOv5 and Faster R-CNN with 95% precision and real-time performance (50ms). The solution included a mobile app for live detection optimized with ONNX Runtime, and extensive image collection and annotation for training. The system was integrated with IoT infrastructure using AES-256 encryption
Results: 95% fewer accidents and 35% lower costs compared to in-house AI development.
Predictive Monitoring System for Transport Vehicles
Client/Industry: Industrial logistics
Problem: Frequent and costly forklift failures causing operational downtime and high service costs.
Goal: Develop an intelligent system for predicting transport vehicle failures based on operational data.
Solution: Our team developed a machine learning module that analyzes vehicle operational data (such as motor hours, number of impacts, load exceedances) and predicts the probability of four types of failures: forks, carriage, engine, and pump. The implementation included synthetic dataset generation, preprocessing, algorithm implementation , and containerized deployment with high-availability HTTPS endpoints.
Problem: Frequent and costly forklift failures causing operational downtime and high service costs.
Goal: Develop an intelligent system for predicting transport vehicle failures based on operational data.
Solution: Our team developed a machine learning module that analyzes vehicle operational data (such as motor hours, number of impacts, load exceedances) and predicts the probability of four types of failures: forks, carriage, engine, and pump. The implementation included synthetic dataset generation, preprocessing, algorithm implementation , and containerized deployment with high-availability HTTPS endpoints.
Results:
- Real-time response: less than 20 milliseconds
- Prediction accuracy: over 91% for forks, carriage, engine, and pump failures
- Downtime reduction: 30–50% less unplanned downtime
- Maintenance planning: up to 2 weeks before predicted failure
- Operational savings: up to €120,000 per year for a fleet of 50 forklifts
- Scalability: supports over 1,000 vehicles with containerized AWS infrastructure
Distributed LLM Locally
Client/Industry: Data-sensitive organizations/insurance
Problem: Need for advanced language model capabilities without compromising data security by sending it to external servers.
Goal: Create a secure solution for utilizing large language models in a local environment.
Solution: We created a secure solution enabling the use of large language models in a local environment, eliminating the need to send data to external servers. The implementation is based on modern RAG technologies, OpenAI, and LangChain, providing full control over processed data and high performance while maintaining complete data privacy.
Problem: Need for advanced language model capabilities without compromising data security by sending it to external servers.
Goal: Create a secure solution for utilizing large language models in a local environment.
Solution: We created a secure solution enabling the use of large language models in a local environment, eliminating the need to send data to external servers. The implementation is based on modern RAG technologies, OpenAI, and LangChain, providing full control over processed data and high performance while maintaining complete data privacy.
Results:
- 100% of data processed locally, ensuring no data leaves the client’s infrastructure,
- 30% annual cost savings on data processing,
- compliance with GDPR and internal security policies, verified through an IT security audit.
Contact
Reach out to us directly and our team will respond to your inquiry as soon as possible.
contact@scientiq.ai
Address
Scientiq
Av. T. Kościuszki 80/82,
90-437 Lodz, Poland
Phone
tel. +48 42 209 18 90