Replify works best when you set up the operating model before inviting the whole team. The goal is simple: every supported customer interaction should have context, one accountable owner, and a clear next step.
1. Complete workspace setup
Start with your workspace name, industry, and primary goal. Replify uses these choices to recommend contact fields, tags, pipeline stages, starter automations, and AI behavior. Review every recommendation before applying it; presets are a starting point, not a locked configuration.
2. Connect one real channel
Open channel integrations and connect the channel your team will operate first. WhatsApp is the primary messaging channel, while Replify also supports Instagram, Messenger, X, LinkedIn, TikTok, YouTube, Google Business, Pinterest, Reddit, Telegram, and Snapchat. Available messaging, comments, publishing, and analytics depend on each network's permissions.
Use the channel connection guide before connecting production accounts.
3. Decide who answers inbound messages
Choose one response mode in AI Agent:
- AI replies only: AI answers unclaimed messages; content-triggered customer replies from automations are paused.
- Automations only: an automation can answer; otherwise the message moves to the team queue.
- Both — automation priority: a guided reply or response-claiming automation goes first, then AI handles what remains.
An active human assignment always has priority. Read how AI, automations, and flows work together before enabling customer-facing replies.
4. Add grounded AI knowledge
In AI Knowledge, add your website, approved files, or manual knowledge. Include the facts customers actually ask about: products, services, locations, hours, delivery rules, returns, booking policies, and escalation boundaries. Test representative Arabic and English questions before activation.
5. Configure the human queue
Set auto-assignment, first-response SLA, overdue reassignment, and admin escalation under inbox operations. Decide which teammates are available and confirm that every shift has an accountable owner. See shared inbox ownership and SLA.
6. Connect work to outcomes
Create the customer fields and pipeline stages your team will actually maintain. Add appointment types if your business books time. Activate only the starter automations that have a clear owner and measurable outcome.
7. Run a controlled launch
Send test messages that cover a known FAQ, a missing-detail question, a human request, an automation keyword, an appointment request, and a case the AI should not answer. Confirm the response, owner, case status, and next step for each one.
Launch checklist
- A production channel is connected and receiving messages.
- Response mode is intentional and understood by the team.
- AI knowledge is approved and tested.
- Human assignment and SLA rules are configured.
- At least one admin can receive escalations.
- Customer-facing automations were tested with safe contacts.
- Every teammate knows how to take over and return work to AI.
Still need help?
Tell us what you were trying to do, the channel involved, and what you expected to happen.