AI assistant for more efficient customer service and product support

LINET struggled with fragmented data and manual transcription of information between systems. This led to delays, errors, and overburdened helpdesk and product teams. Efficient responses to customer inquiries and preparation of technical documentation were challenging and slow.

A comprehensive AI solution was developed, combining a helpdesk assistant, product team support, and system integrations. The AI assistant generates response drafts to customer inquiries using LLMs (GPT), integrates with Salesforce, SAP, and other data sources. Product specialists use a hybrid search combining full-text and vector embeddings for documentation retrieval. The solution also integrates business processes “from quote to invoicing,” eliminating manual work and reducing error rates. The entire system runs on Azure, leverages Entra ID, and is fully auditable.

Customer service operators (helpdesk).
Product specialists.
Sales and proposal teams.

6

.NET 6 / C#, Azure App Service, Azure Functions, PostgreSQL + pgvector, Azure OpenAI (GPT-4o, embedding-ada-002), Semantic Kernel (Microsoft), Salesforce (REST/SOAP integration), Azure DevOps CI/CD, Application Insights, Log Analytics, ASP.NET Razor Pages, RTK Query, Axios.

Head of business unit

Full implementation

Operations and maintenance are managed by LINET’s IT team.

Salesforce admin.
Internal IT support (for infrastructure, permissions, SAP).
Product specialist (for output verification and feedback).

Azure subscription with access to OpenAI services and databases.
Integration with internal SAP, K2, Salesforce.
Access to documentation and catalog data.
Costs include Azure services and vendor SLA support.

• Significant acceleration and automation of work for customer service and product specialists.
  
• Reduced administrative workload thanks to end-to-end process automation from quotes to invoicing.
  
• Lower error rates and faster response times to customer requests.
  
• Increased credibility and transparency thanks to direct references to source documents.

LINET, through collaboration with Cloudfield, significantly streamlined customer service and internal business processes. The AI assistant–based solution unified data, accelerated response times, and reduced administrative burden across the organization.

Top-quality AI integration into existing infrastructure. Close cooperation with the client and iterative development. Transparent access to data ensuring output verifiability.

Inconsistent data in SAP (duplicate BOM, missing names). Need for explainability of AI outputs. High demand for accuracy and verification in live operations.

Start with a narrow use case with clearly measurable benefits and validated data sources. Focus on continuous testing of AI outputs and transparent interpretation of responses.