When AI
sposta the numbers.
A selection of projects released to production across SMEs and large groups. Each case reports verifiable KPIs, technology stack and actual duration. All clients appear in anonymised form under confidentiality agreements.
Infrastructure
scalable.
Demand forecasting and inventory optimisation
ML models on AWS for short and medium-term demand forecasting. Reduction of stock-outs and tied-up capital, data pipeline on Airflow.
Document intelligence for cases
LLM + OCR pipeline for structured extraction of clauses, amounts and deadlines from contracts and policies. Core banking integration.
Network asset predictive maintenance
Predictive models on distributed IoT sensors to anticipate failures in electrical grid assets. Reduction of unplanned interventions and operating costs.
Real-time transaction fraud detection
ML model on AWS Lambda for real-time scoring of every transaction. Latency under 20ms, integrated with existing core banking without architectural changes.
Multi-site data lake on AWS
Data centralisation from 8 plants on AWS S3 + Glue. Analytics on Athena, real-time operational KPIs on a Grafana dashboard, elimination of information silos.
Customer churn prediction
ML model on AWS SageMaker to identify at-risk customers 30 days in advance. Automatic trigger of retention campaigns, integrated with CRM.
Automatic classification of logistics documents
OCR + LLM pipeline on AWS to automatically classify and archive invoices, delivery notes and waybills. Zero human intervention on standard documents.
Plant energy consumption forecasting
LSTM model on OCI for 24–72h energy consumption forecasting. Production scheduling optimisation on favourable time slots, peak reduction.
Document QA pipeline on OCI
Automation of GMP regulatory document and batch record compliance checks with fine-tuned LLM. Integration with LIMS system, complete audit trail.
Automatic generation of analysis reports
LLM pipeline on AWS to transform structured data from CRM and ERP into personalised narrative reports for management. Layout and tone configurable per client.
Multi-entity financial consolidation and forecasting
Data pipeline on OCI for consolidating the accounts of 6 companies. ML model for cash flow forecasting and automatic alerts on treasury anomalies.
Dynamic pricing on product catalogue
ML model on AWS for real-time price optimisation across 50k SKUs. Automated competitor monitoring and repricing driven by demand elasticity.
Fleet scheduling optimisation
Combinatorial model on AWS for automatic assignment of vehicles and drivers to daily missions. Integration with existing TMS, including contractual constraints.
MLOps pipeline su AWS SageMaker
Complete MLOps infrastructure: automated training, model versioning, drift monitoring, automatic redeployment when performance thresholds are exceeded.
Legacy ERP migration to OCI
ERP infrastructure migration from on-premise to Oracle Cloud with zero downtime. Architecture optimisation, automated disaster recovery, infrastructure costs halved.
RAG on B2B catalogue and price lists
RAG system on AWS to allow the sales network to query catalogue, availability and prices in natural language. Integrated with ERP, response in seconds.
Predictive returns analysis by category
ML model on AWS for forecasting return rates by category, season and sales channel. Input for purchasing, stock allocation and refund policies.
Sentiment analysis on product feedback
AWS Comprehend pipeline + fine-tuned model for analysing thousands of reviews per day across multiple marketplaces. Automatic alert on emerging negative trends.
Photovoltaic plant production optimisation
ML model on OCI for real-time optimisation of inverter configuration and solar tracking. Integration with existing SCADA, no hardware replacement.
ETL pipeline for multi-site ERP consolidation
AWS Glue + Airflow for data consolidation from the ERPs of 4 plants with heterogeneous systems. A single harmonised data model for analytics and reporting.
Zero data
leaving.
Visual quality control on line
AI Vision with Nvidia Jetson Orin for defect detection in the production line. On-edge inference at 60fps, integrated with existing PLCs and MES. Zero data to cloud.
Private AI assistant for law firm
On-prem LLM on Sovereign AI for querying contracts, judgments and internal firm documents. Zero data egress, native GDPR compliance, no cloud vendor.
AI supervision on extrusion line
Sovereign AI with vision module on an extrusion line. Geometric and chromatic defect detection at 70fps, PLC alert in under 5ms, direct MES integration.
Predictive maintenance on industrial presses
Sovereign AI acquires data from accelerometers and thermal sensors on 8 presses. On-device anomaly detection, contextualised Telegram alert before the failure.
AI assistant for sales network
Sovereign AI at head office: sales agents query availability, technical sheets and customer history via chat. Integrated with ERP, zero cloud, 45 active users.
Batch record analysis on Sovereign
On-prem LLM for extraction and compliance verification on GMP batch records. No production data leaves the plant. Complete on-device audit trail.
Multi-site Sovereign fleet
3 Sovereign nodes distributed across 3 plants, managed with balenaOS. Centralised OTA updates, unified fleet monitoring, zero-touch configuration.
AI assistant on tax and accounting regulations
Sovereign AI for an accounting firm: RAG on up-to-date tax regulations, Revenue Agency circulars and civil code. Corpus updated automatically.
Knowledge management on technical documentation
On-prem LLM trained on technical drawings, BOMs and maintenance manuals. Technicians query in natural language, including from mobile on the internal network.
Air-gapped Sovereign AI for public body
Sovereign AI deployment on a classified network. LLM for searching internal document archives, no cloud vendor involved, public administration regulatory compliance.
Processes that
manage themselves.
Returns management automation from sales network
Custom AI Agent that automates the management of returns from the sales network to head office, integrated with existing ERP and CRM. Operator escalation only for edge cases.
CRM agent on Telegram
AI Agent on Telegram for the commercial team: consults the CRM, updates opportunities, logs post-call notes and generates the daily report. All without opening the management system.
Intelligent production alerting via Telegram
MQTT pipeline → anomaly detection → contextualised Telegram notification. The operator receives the probable cause and recommended action, not just the out-of-threshold value.
B2B customer care on WhatsApp
AI Agent on WhatsApp for managing B2B customer requests: order status, documents, complaints. Operator escalation for complex cases, 24/7 operation.
Client onboarding automation
n8n + LLM pipeline for document collection, KYC verification and contract generation. From first contact to digital signature in 48h, without manual intervention.
AI agent for inbound lead qualification
Agent that analyses every inbound lead, enriches the profile with public data, assigns a score and routes it to the correct salesperson. Zero manual intervention.
Automated operational reporting
Prefect + LLM pipeline that every Monday automatically generates the weekly operational report from ERP and MES data, with narrative commentary and highlighted anomalies.
IT ticket triage and classification
AI agent integrated on Slack that classifies every new ticket, assigns priority, suggests a solution from the knowledge base and automatically notifies the correct team.
Order agent on Telegram for sales network
Agents send orders on Telegram in natural language. The AI interprets, checks availability on ERP and automatically confirms the order. Zero entry errors.
Automated competitive monitoring
n8n + LLM pipeline for daily analysis of competitor prices across multiple marketplaces. Automatic report every morning with alerts on significant changes.
Could be
essere il your.
Free discovery call to frame your problem and understand whether the SOlutica approach can work in your context.