AI Assistant and System Integration for LINET: Digitizing Customer Support and Business Processes
About The Customer
LINET spol. s r.o. is a leading Czech company and one of the world’s top manufacturers of hospital beds. Since its founding, the company has focused on innovation, quality, and design to deliver highly functional products and exceptional care to clients worldwide.
Challenge
LINET faced challenges related to fragmented information and documentation, which slowed down their ability to deliver fast and effective customer service. Support teams and product specialists often had to manually transfer data between systems, resulting in errors, administrative burden, and inefficient workflows.
Requirements
LINET sought to implement AI technologies that would:
- Unify data and enable fast, accurate, and automated responses.
- Streamline support desk operations.
- Make it easier for product specialists to prepare technical documentation and responses.
- Integrate existing information systems (CRM, ERP, billing) and automate the business process from offer to invoice.
Why Cloudfield?
Cloudfield brought:
- Deep expertise in AI and document processing.
- A flexible approach tailored to LINET’s specific operational needs.
- Robust integration capabilities with existing IT systems, including Salesforce.
- A transparent solution with clear references to source data, eliminating the “black box” effect.
Implementation
The solution was structured into three core components:
1. AI asistent for helpdesk
A modern multi-agent system integrated into Salesforce, designed to generate pre-filled response suggestions for helpdesk operators:
- A process agent composed of three modules:
- Query analysis using LLMs – breaks down incoming messages into sub-questions and verifies the relevance of available sources.
- Key information extraction – specialized agents query both structured and historical databases, using parameterized search and semantic indexing.
- Final answer synthesis – merges selected candidates into a single, coherent response.
- Transparent UI - operators see search parameters and candidate snippets, enabling quick validation without relying on a “black box”.
- Clean API interface – the module can be called from other systems (web, mobile, internal portal), ensuring reusability of the solution.
2. Support System for Product Teams
A unified platform combining various types of documentation (manuals, test protocols, certificates, etc.) to streamline responses to expert inquiries:
- Combines traditional full-text with vector-based (semantic) search powered by embeddings to capture questions with different wording.
- Provides contextual references – each answer includes quoted passages and source documents to enhance credibility.
- Integrated workflows save time for product specialists preparing materials for tenders and marketing.
3. Integration of Existing Information Systems – “Offer to Order”
Consolidation of heterogeneous sources made it possible to automate the entire process from offer to invoicing:
- Integration with CRM, ERP, and billing modules eliminated manual data re-entry and shortened the order processing cycle.
- Data normalization reduced error rates and increased transparency in key performance indicators (lead time, cash flow).
Outcome and Benefit
- Significant acceleration and automation of work for both customer service and product specialists.
- Reduced administrative burden through automation of the offer-to-invoice process.
- Greatly reduced error rate and faster response to customer requests.
- Increased credibility and transparency through direct links to source documents.