Documentation Index
Fetch the complete documentation index at: https://docs.mrassistant.ai/llms.txt
Use this file to discover all available pages before exploring further.
This quickstart is conceptual: it describes the order of decisions most teams make. For field-level request shapes, use the API reference when your team is ready for implementation detail.
1. Clarify the consumer journey
Decide what the customer should accomplish in one session: get an answer, complete a task, or reach a human. That drives how you structure agents, languages, and handoffs.
2. Align voice and channels
Choose where the experience lives first (web, app, phone, or a mix). Realtime voice usually involves coordination between your product surface and your backend; see Web and voice.
3. Plan integrations deliberately
For B2C scale, teams often introduce a small set of well-defined actions (lookups, bookings, tickets) behind a stable interface rather than growing ad hoc scripts inside prompts. MCP for B2C explains why that pattern is common.
Think in terms of instructions, languages, voice persona, and what the assistant is allowed to do. Iterate with real callers and transcripts rather than only desktop testing.
5. Roll out in phases
Pilot with a narrow audience, measure containment and satisfaction, then widen traffic. Keep operational playbooks (escalation, content updates, incident response) next to the technical rollout.
Next steps
