Dental practice / Instagram DM automation
From Instagram DMs to a reviewable workflow
I connected a dental practice's Instagram conversations with its CRM, using AI to interpret enquiries and a defined workflow to manage replies, records and staff handoff.
My role: workflow design, implementation, integration testing and operating documentation. Delivered in April 2026.
- 01 / ReceiveInstagram + ManyChat
Incoming messages and button selections.
- 02 / Coordinaten8n + Claude
Check pause state, load context and interpret the enquiry.
Validate the response before sending - 03 / ReplyBack through ManyChat
A reply, relevant options or an image, delivered to Instagram.
Alongside the reply, keep the conversation connected
GoHighLevelContact, intent tags, conversation notes and staff pause control.
Google SheetsConversation history for context and review.
Simplified workflow illustration.
The conversation was only the beginning
An Instagram enquiry could move from a general question to interest in a consultation, or a request to speak with the team. Replies needed context from earlier messages. Staff needed the conversation connected to a contact record, with a clear way to take over.
I built around the existing tools: ManyChat for the Instagram connection, n8n to coordinate the workflow, and GoHighLevel for the team's contact records. Google Sheets supplied recent conversation history and a log for review.
Give the model a defined job
Claude interpreted intent and returned structured fields for the reply and next step. Code checked those fields, guarded against raw model output reaching the conversation, and suppressed repeated options or images. Practical rules stayed in code where I could inspect them.
A closer look / Staff control
Make room for a person.
A request for the team triggered a handoff reply and set the contact's pause flag. A direct phrase check backed up the model's classification.
Staff could also pause through GoHighLevel or the ManyChat inbox. The operating guide explained how to resume, where records appeared and what to check if the Instagram connection needed refreshing.
Incoming DMI'd like to speak with the team.
Handoff replyI'll pass this to the team.
- Automation
- Paused
- Conversation
- Note appended
What I delivered
- The DM workflow: contextual replies, mid-conversation options, consultation links, relevant image sending and typing indicators.
- The staff workflow: CRM contact sync, intent tags, conversation notes, pause controls and a separate conversation log.
- The handover: an operating guide and delivery records. April live smoke tests covered replies, buttons, images, typing indicators, CRM integration and logging.
Follow-up campaigns, a reporting dashboard and automatic pause when staff replied directly in the Instagram app remained proposed extensions.
Phase 1 was delivered in April 2026. Automated replies were later paused at the client's request in June.
The operational lesson
The entry point deserves as much attention as the reply. Later feedback exposed a mismatch in the fixed welcome reply used for ad traffic, before messages reached the language model. It reinforced a practical rule for my work: test each way a conversation starts, then test how staff take it over.
Have a conversation workflow to connect?
Tell me where enquiries arrive, which tools your team uses and when a person needs to step in. I'll help you find a sensible scope for a custom system.
Discuss your workflow