FA©26

Case study — 2026

NDA · Voice AI client

Voice AI Follow-up Workflows

Role
Automation · n8n · GoHighLevel
Status
Delivered to the client
In one line
The n8n workflows behind an AI voice agent, writing back to GHL.

n8n workflows behind an AI voice agent connected to GoHighLevel: onboarding intake, tenant lookup, knowledge-base sync, in-call tools, a post-call workflow that writes back to GHL, and an error handler.

client build · under NDA
n8n editor showing a weekly agent-insights workflow: a Monday schedule loops over accounts, checks call volume, asks Claude for findings, stores them and posts a summary to Slack. Request URLs are blurred.
The weekly insights workflow in n8n. Request URLs blurred.

01 — The problem

A voice agent is only as useful as what happens around the call: knowing who is calling, what the agent is allowed to say, and whether the outcome ever reaches the CRM. Without that plumbing, every call ends with someone retyping notes.

The agent also needed a way to get better: someone had to read the week's calls and say what went wrong, across every account.

Client work under NDA. The system is described in full; the names stay private.

02 — What I built

  • Onboarding intake

    Captures and routes new callers.

  • In-call tools

    Tools the voice agent calls during a conversation.

  • Post-call workflow

    Writes call outcomes back to GoHighLevel.

  • Custom-value sync

    Keeps GHL custom values and the voice agent's knowledge in sync.

  • Data clean-up scripts

    Python scripts for field consolidation and duplicate surveys.

  • Weekly agent insights

    Every Monday the workflow loops over each account, skips low-volume ones, has Claude write findings from the week's calls, stores them and posts a summary to Slack.

03 — How it runs

The loop — Caller onboarded, then Agent looks up the account, then In-call tools answer, then Outcome written to GHL, then Weekly findings by Claude, then Knowledge kept in sync.
  1. Caller onboarded
  2. Agent looks up the account
  3. In-call tools answer
  4. Outcome written to GHL
  5. Weekly findings by Claude
  6. Knowledge kept in sync

04 — Proof

  • Error-handling workflow

Honest note

Described from the workflows themselves; the client, its product and the workflow files stay private.

05 — Stack

  • n8n
  • GoHighLevel
  • Retell
  • Conversation AI
  • Claude
  • Python
  • SQL

Next project

Scheduled GoHighLevel Sync Backend

Keeps GoHighLevel and a Lightspeed retail system in step.