Case Study · Support Germany

18 years of support history, answered in minutes

A globally active industrial client centralized its support teams. Ragsense turned 18 years of German support emails, attachments and SharePoint documents into one AI knowledge base that the whole support team can query in French, English or German.

Support KB · Internal workspace
The Support Knowledge Base home page: over 75,000 emails processed, 15,000 conversation threads, 6,800 entities extracted, in German, English and French

From fragmented archives to one intelligent knowledge base

Years of support knowledge were scattered across email archives and several SharePoint sites. Ragsense developed and operates an AI-powered retrieval system that unifies them into one knowledge base for the global support team. A dashboard reports on the support topics handled, and dedicated filters let agents dig into previous cases to speed up every resolution.

Support KB · Email Archive Insights
The Email Archive Insights dashboard: message volume over 18 years, conversation topic clusters, message categories and languages

Hybrid search across the complete support history

The agent combines full-text keyword precision with semantic understanding, merged by Reciprocal Rank Fusion. Support agents ask in French or English and instantly get answers from German sources, with the full conversation context and links to the original emails and documents. The project started with a five-week pilot on the email archives, the core of the support knowledge, before expanding to further sources in weekly agile sprints.

Extracting entities

Understanding and grouping entities is key

The same customer, product or feature is named in many different ways across years of emails. In a preparatory phase, we built an AI-powered workflow that identifies and clusters these variations, so the agent can reliably find every support case of a given customer or product.

  • 2,405 customers identified in the support files
  • 6,856 products and features extracted and indexed
  • An entities viewer to validate the quality of the extraction
The Extracted Entities viewer: 2,405 customers and 6,856 products, with name variations grouped into clusters
70,000+

Support emails indexed

257,000

Attachments searchable

18

Years of support history

3

Working languages

AI overviews

Translating support history for a centralized follow-up

International corporations increasingly centralize their support teams, who then have to follow up cases recorded in a language they don’t use natively. We analyze, translate and compress every support issue into indexed AI overviews, making the whole archive searchable at lightning speed.

  • Questions in French or English, answers from German sources
  • An AI summary for every message, next to the original
  • Resolution times cut from days to minutes
The Support Assistant answering a technical question in English from German emails, with the source message, its AI summary and the full thread
Three search modes

Select how deep the agent should search

Support agents choose the depth of research according to the time available and the complexity of the case.

  • Standard: fast, focused answers for straightforward questions
  • Thorough: a wider net for complex questions, including attachment summaries
  • Deep Research: maximum coverage with multi-angle queries across every field and the full attachment text
The three search modes, Standard, Thorough and Deep Research, with their result counts, quality thresholds and search fields
Email & SharePoint browser

Every email, attachment and document, one click away

Beyond the assistant, the platform allows a deep search in email conversations, attachments and the connected SharePoint sites that hold technical data and support documents.

  • Full conversation threads, filtered and expanded in place
  • Hybrid search across SharePoint documents
  • Saved conversations for admins to audit
The workspace modules: Messages, AI Assistant, Statistics, Attachment Browser, Extracted Entities and Conversation History
Built-in feedback

Real-time feedback from the support team

Support cases in an industrial corporation are complex. Constant feedback from the support team using the tool helps the AI agent improve the quality of its answers and fine-tune the support scenarios.

A thumbs-down on an answer opens a comment box asking what could be improved

Ready to see it in action?

Schedule a personalized demo and discover how Ragsense can transform the way your organization works with AI.

Request a Demo More Case Studies