Back to projects

XeroWA AI / WhatsAI Assistant

I am building XeroWA under Aviro Technologies, a DPIIT/MSME-registered AI startup. It is a live WhatsApp-first product for Indian SMBs that handles approved replies, lead qualification, appointments, reminders, follow-ups, broadcasts and owner handoff.

Live product site GitHub repo

WhatsApp lead work in one system

Indian businesses receive customer questions and sales leads on WhatsApp. XeroWA gives the team approved replies, qualification steps, appointment capture, follow-ups and an owner handoff path inside one controlled workflow.

My contribution

I own the product, engineering and operating workflow: landing site, dashboard, integration boundaries, tenant data model, Aviro startup positioning and verification checks.

Technology

Next.jsTypeScriptSupabase / PostgreSQLMeta WhatsApp Cloud APIVercel serverless routesVitest

Message to owner handoff

  1. A customer sends a WhatsApp message to the business number.
  2. The Meta webhook verifies the request and routes it to the correct tenant.
  3. The tenant playbook selects an approved reply or qualification path.
  4. The system stores the conversation, lead, appointment and follow-up state in Supabase.
  5. The dashboard shows conversations that need owner action.

Checks around tenant data

  • Meta webhooks receive and route inbound WhatsApp messages.
  • Business-approved playbooks control customer replies and lead qualification.
  • Supabase and PostgreSQL store tenant, conversation and lead data.
  • Readiness checks, RLS tests and keyword-engine tests verify core paths.

Approved replies and structured automation

Tenant playbooks store the replies that customers receive. The system uses those playbooks for qualification, appointment capture, follow-ups and handoff.

Aviro Technologies context

XeroWA is being operated under Aviro Technologies Pvt. Ltd., a DPIIT/MSME-registered AI startup focused on practical workflow automation for Indian businesses. I keep the product positioning intentionally controlled: approved replies first, visible owner handoff, and business-specific playbooks instead of unpredictable customer-facing AI responses.