All sectors

Measuring adoption and analysing conversations

Administrators track actual adoption of the assistant by segment and receive an AI analysis of conversations: topics, pain points, missing content. The content to enrich first becomes visible.

Talk to an engineer

Illustrative mock-up: the names, values and documents shown are fictitious.

Context

An assistant deployed at scale whose actual use remained poorly understood.

The problem

Knowing who uses the tool, for what, and what is missing from the knowledge base.

What we built

A user event log, adoption metrics and segments, active user curves, and an AI analysis triggered by the administrator over a given period: themes, sentiment, pain points, missing content, and an actionable report.

Steps

01Instrumentation of key events
02Adoption dashboards
03Batch AI analysis of conversations
04Feedback loop to content owners

Hosting and models

Client's Azure cloud (West Europe): API and workers in managed containers, PostgreSQL with vector search, object storage. Open embedding and reranking models (Qwen3 family) served by vLLM on dedicated GPU machines. Commercial large models (Gemini, Claude) called via a LiteLLM proxy, which avoids dependence on a single provider.

Services involved

AI in your applications

We name a client only with their written agreement. The budgets and detailed results of our engagements remain confidential.