Case studies
At a major Italian bank, an LLM agent sorts 1M+ help desk tickets a year into 400+ categories: 81% accuracy vs 75% human, resolution time halved.
A major Italian bank cut handling time on inheritance document checks by 80% with Azure Document Intelligence and an LLM. Exceptions go to operators.
OCR + LLM pipeline on FastAPI and Kafka (16 workers, OpenShift): 2,500+ financial documents a day, ~30 checks on each, 84% less manual work.
We rewrote the AI pipelines of a consumer finance platform: fewer redundant calls, tighter context, spend caps, rate limits. AI costs fell by about 80%.
We rebuilt retrieval for Estyl's fashion AI assistant with hybrid vector search, structured filters and re-ranking: 90%+ of responses fully correct.
A GDPR-aware real-time voice agent: replies in under 1s, data kept separate per clinic. First month at a 40+ doctor clinic: 0 missed calls.
The platform behind IGEA's voice agent: per-clinic settings, doctors' calendars and hours, appointments, call logs, SMS reminders and outbound calls.