Botpress Pilot Client Agent Deployment (Onboarding)

Botpress Pilot Client Agent Deployment (Onboarding)

1. 01

   Create the client workspace in Botpress and confirm the plan tier against expected conversation volume

   The Free plan caps at 100 conversations, 3 seats, and 3 AI agents. Plus at $150/mo includes 250 conversations/mo and unlimited AI agents, so a pilot client with heavy ticket flow will breach the Free cap inside the first month.

2. 02

   Open Agent Studio and map the pilot client's ticket types before building anything

   List the refund, account update, and multi-step workflows the agent must close end-to-end. Anything the agent cannot resolve should be flagged now as an escalation path rather than discovered in production.

3. 03

   Train the agent on the client knowledge base and set tone rules

   Load the client's FAQ, policy documents, and pricing into the knowledge base, then test against the five most common ticket types. Verify the agent answers in the client's voice, not a generic support voice.

4. 04

   Connect the helpdesk and channel integrations

   Wire HubSpot, Notion, and Jira where the client already works, then attach the messaging channels in scope (WhatsApp, Instagram, Slack, Messenger, email, webchat). Confirm each channel routes into the same agent rather than spawning duplicate flows.

5. 05

   Configure escalation handoff with full conversation history

   Set the conditions that push a ticket to a human (refund above a threshold, account changes requiring verification, frustrated sentiment) and confirm the human agent receives the transcript, not a blank ticket.

6. 06

   Apply white-label webchat and custom domain Help Center on the Plus plan

   Replace Botpress branding on the webchat widget and point the Help Center at the client's domain. This is what lets the agency present the deployment as its own support product on a retainer.

7. 07

   Run a two-week monitored pilot and log resolution rate per ticket type

   Track which workflows close without human touch and which escalate. Use the conversation logs to retrain weak flows before the pilot converts to a paid managed service tier.