3

Enterprise Internal Assistant for Global QSR

Internal conversational assistant for a global QSR brand (~36k staff): cited answers from procedures and glossary; persistent chat, saved questions, searchable glossary; PWA; multi-language (PL, EN, UK); FastAPI, Next.js, Chakra UI.

Internal conversational assistant for a global quick-service restaurant brand (~36k staff). Staff navigate procedures, terminology, and documents via natural-language questions and cited answers; the system consolidates access to internal knowledge (procedures, glossaries, policy documents) and the company intranet.

Purpose

  • Business: Instant, grounded answers from internal knowledge; standardise access to company terminology and policies; support multiple languages (PL, EN, UK) for diverse, multilingual teams.
  • Technical: Conversational AI over enterprise content with cited sources (RAG-style, managed AI/search); persistent chat history, saved questions, searchable glossary backed by cloud database and REST APIs; PWA for mobile-first, installable use in the field; authentication (identity provider) and backend auth so only authorised users access the app and APIs.

Stack

  • Backend: FastAPI, managed SQL (e.g. Cloud SQL), managed AI/search services; deployed on Cloud Run.
  • Frontend: React/Next.js, TypeScript, Chakra UI; responsive, PWA, multi-language, accessible (ARIA, keyboard navigation).

Challenges

  • Integrating conversational AI with existing auth, session handling, and UI (chat, history, ratings) so answers, sources, and feedback flow correctly end-to-end.
  • Unifying multiple content types (glossary, documents, procedures) into one search/answer experience with reliable citations.
  • Mobile UX: performance, scroll behaviour, perceived speed; data model and API design for saved questions, chat history, and ratings with consistent error handling.
  • i18n and brand/UX guidelines.

Outcome

For users: a single place to ask questions and get answers with sources, reuse saved questions, and look up terminology in their language. For the organisation: standardised access to policies and terminology; auditable usage via chat history and ratings; scalable cloud deployment. Full-cycle build of an internal AI assistant for a large QSR brand, from design and integration to deployment and UX refinement.