Two Reference Delivery Stacks
For any client build, you’re choosing between two architectures. Pick per client; don’t dogmatically use one.
Stack A — “All-in-one” (GoHighLevel-centered)
For: local/service SMBs who want one login, aren’t technical, and value simplicity over cost.
┌────────────────────────── GoHighLevel ──────────────────────────┐
Inbound → │ CRM · Pipeline · Calendar · Missed-call text-back · Reviews · │ → Owner
(calls, │ Email/SMS workflows · Forms · Reporting │ (approves)
forms) └───────────────────────────────┬──────────────────────────────────┘
│ (webhook)
Claude skill (drafts replies / quotes)
- Pros: fast to stand up, one bill, client can run it after handoff, A2P handled in-platform.
- Cons: monthly platform cost, less flexible, some lock-in.
- Add-ons: Retell.ai for voice; cal.com if they prefer it over GHL calendars.
Stack B — “Composable” (n8n-centered)
For: clients who are technical, cost-sensitive at scale, already have tools they like, or need data control.
Inbound (forms, email, calls, CRM events)
│
▼
n8n ──► Claude (draft/analyze/summarize) ──► action
│ (SMS via Twilio, email,
├──► Twilio (SMS / missed-call) calendar write, CRM update)
├──► EspoCRM / HubSpot (records)
├──► cal.com (booking)
└──► Metabase (dashboards)
- Pros: cheap at scale, no lock-in, full control, self-hostable (data stays with client).
- Cons: more setup, you own the wiring, client needs J&M (or the retainer) to maintain it — which is a feature for recurring revenue.
How to choose (decision rule)
- Non-technical local SMB, wants simple, one bill? → Stack A.
- Technical, cost-sensitive, data-control needs, or already tooled up? → Stack B.
- Either way: J&M’s brain layer is always Claude, the client hub is always Notion, and SMS always goes through A2P 10DLC + consent.
The stack is a means, not the offer. Clients buy hours and dollars recovered — an operating system they own — not GoHighLevel vs. n8n. Choose the stack that ships their fixes fastest and that they can live with.