Messaging Transaction Assistant Starter

This starter demonstrates a compact transaction flow inside a messaging-style assistant. The motivating signals are Meta's April 2026 prepaid-recharge flow and the May 2026 Business AI on WhatsApp launch for Indian small businesses, but the implementation is repo-native and generic: it teaches bounded business-context grounding, intent capture, plan selection, user confirmation, human takeover, and payment handoff boundaries without copying a vendor UI.

Status

starter

What It Demonstrates

• keep a small business context grounded in local profile, catalog, and support notes
• capture a user's transaction intent from a chat message
• select a simple plan from local fixtures
• keep human takeover available when the business wants to step in directly
• require explicit confirmation before payment handoff
• keep payment execution outside the assistant
• record the source lineage that inspired the starter

Quick Start

Run the repository-level smoke check:

python3 scripts/verify_example_projects.py

Or inspect the starter directly:

python3 ecosystem/examples/messaging-transaction-assistant-starter/src/run_demo.py

The demo prints the business context sources that the assistant is allowed to use before it proposes a payment handoff.

Current Product Signal

Meta's current Business AI on WhatsApp launch makes one reusable lesson more concrete: the business-side assistant should answer from a bounded business profile, catalog, and support context, while leaving seller override and payment execution outside the automation loop.

Boundaries

This starter does not process payments, store card data, call telecom APIs, or send real messages. It also does not imply a live WhatsApp Business Platform integration. The assistant stops at a structured handoff that a real business operator or payment surface would need to own.