Work · Case study 01 of 18 · AppPilot
AppPilot overview
AI Engineering

AppPilot

Conversational AI for operational workflows

SectorHR, hospitality and banking
RoleAI Engineering
StackRAG, LLM tool calling, Document OCR, Voice ordering, Node.js, React

The challenge

Most internal questions are easy but expensive. Leave balances, outpatient limits and claim rules already live in policy documents, yet every answer costs HR ten minutes, and the claim that follows still needs forms, receipts and manual checks.

What we built

  • Policy documents are ingested and retrieved at question time, so every answer is grounded in the employer's own rules.
  • Claims run as a guided conversation: the assistant asks only for what is missing and accepts receipts as uploads.
  • OCR extracts clinic, doctor, date and amount, reconciles them against the declared figure and flags mismatches before a ticket is raised.
  • The same engine was redeployed as a voice ordering assistant for restaurants and a bilingual English and Urdu banking assistant.

Outcome

3 production deploymentsEvery query logged for auditHR-maintained knowledge base
AppPilot screen 1
AppPilot screen 2
AppPilot screen 3
AppPilot screen 4
AppPilot screen 5
AppPilot screen 6