B4
Track B — Multi-Agent Orchestration & Decision Support

Product Compliance Advisor Agent

Systems where multiple specialized agents collaborate, hand off tasks, and reach a combined decision.

Background

Small manufacturers often don't know which BIS (Indian Standards) certification applies to their product, leading to non-compliance and penalties. The BIS portal is hard to search without already knowing the exact standard code.

Objective

Develop a conversational agent that asks clarifying questions about a product and recommends the applicable Indian Standard(s), along with next steps for certification.

Key Features

1. Adaptive clarifying-question flow based on product description 2. Search over a curated standards database 3. Ranked standard recommendation with explanation 4. Compliance summary report with next steps

Expected Solution

A multi-turn clarification agent that narrows down product category A search/recommendation layer over a curated standards database A report generator summarizing applicable standards and next steps

Suggested Technology Stack

Orchestration: LangGraph (multi-turn clarification loop) Search: Keyword + semantic search over a curated standards dataset LLM: For question generation and report writing Data: Curated set of 30–50 common BIS standards (electronics, toys, food packaging, textiles)

Expected Outcomes

Faster identification of applicable compliance standards for small manufacturers Reduced non-compliance and associated penalties A reusable clarification-driven recommendation pattern

Possible Use Cases

A toy manufacturer identifying required safety certification A food packaging startup checking applicable standards An electronics maker verifying certification requirements before launch

Evaluation Parameters

Quality of clarifying questions (do they genuinely narrow down the category?) Accuracy of standard recommendation Clarity of the final compliance report