AI Email Response Automation Builder (Make.com + Claude API + GHL + Instantly.ai)

AI Email Response Automation Builder ( ---------- + Claude API + GHL + Instantly.ai)

We are a reputation management company with a working outbound cold email engine. We are now building the next layer: an AI-powered reply-handling and follow-up automation system for interested prospects. This is a serious technical build. Not a beginner automation job.

If you are hired and you perform, this turns into ongoing paid work building and maintaining the rest of the business automation stack. We have a real backlog. The right person does not leave.

What You Will Build

A production-ready system that:
1. Pulls interested-prospect replies from Instantly.ai and pauses the sequence immediately
2. Fetches the full email thread and sender persona
3. Sends thread context plus business data to Claude API with a structured system prompt
4. Parses Claude JSON output and routes: auto-send, hold in Slack for VA approval, or escalate to founder
5. Sends replies through Instantly.ai preserving thread and sender persona
6. Sends SMS follow-up 15 minutes after email via JustCall or GHL
7. Updates GHL pipeline, tags, and 21 custom fields
8. Triggers stage-specific follow-ups when a prospect goes cold (5 follow-up sequences)
9. Slack approval workflow with approve, edit, and reject actions inside modals
10. Phased rollout from semi-automated to fully automated

Full build spec (v3, 19 pages) shared only with shortlisted candidates. It covers every module, every rule, every test. We already did the thinking. We need execution.

Project Structure

Phase 1: Fixed-price build. 50,000 PHP. Paid per milestone (40/15/15/15/15 percent across 5 milestones).

Phase 2: Paid ongoing maintenance and optimization. Separate paid arrangement. Retainer range 20,000 to 40,000 PHP per month depending on scope and volume of new work.

Phase 3: Long-term role. If you are strong, fast, reliable, and easy to work with, you become the go-to technical builder across the business. We have a real pipeline of future builds.

Bonus: 10,000 PHP for clean, on-time v1 delivery (all 5 milestones passed within the stated timeline).

Required Experience

You should have direct, production experience with most or all of these:
- ---------- or n8n (multi-router, error handlers, data stores)
- APIs, webhooks, HTTP requests, JSON parsing
- Claude or Anthropic API with structured JSON output enforcement
- Instantly.ai API (reply webhooks, thread fetch, reply send)
- GoHighLevel (custom fields, pipelines, workflows, webhooks)
- Slack Block Kit, Interactivity, modals for approval flows
- SMS integrations (JustCall, Twilio, or GHL)
- CRM automation and lifecycle workflows

Who This Is For

You fit if:
- You have built multi-step systems with APIs, routers, and approval flows
- You think through edge cases before they break production
- You move fast, communicate blockers early, and do not disappear
- You can work independently from a detailed build spec

You do not fit if:
- You only know basic Zapier-level automation
- You have never worked directly with APIs, webhooks, JSON, or routing logic
- You need constant hand-holding
- You drag projects out

Speed Matters

The build spec has a 48-hour checkpoint. By end of Day 2 after access is granted, you must demo Modules 1 to 9 working. If you cannot, the project ends and you are paid only for completed milestones. This is non-negotiable.

Timeline target: 7 to 8 days for a clean v1. Tell us your realistic number.

To Apply

Do not send a generic application. Start your application with:

READY TO BUILD

Then answer every question below. Vague answers are an automatic pass.

Screening Questions

1. What is the most advanced automation system you have personally built. Walk through the workflow: inputs, routers, outputs, error handling.
2. Have you built against the Instantly.ai API directly (reply webhook, thread fetch, reply send). If yes, what broke and how did you fix it.
3. Have you used the Claude or Anthropic API inside a live production automation. How do you enforce JSON-only output and recover when the model returns malformed JSON or wraps it in markdown.
4. Have you built Slack approval workflows with approve, edit, and reject actions including modals with pre-filled content. Explain what you built.
5. How do you prevent a webhook from processing the same event twice in ---------- .
6. How do you handle Claude API rate limits and retries without duplicating messages.
7. The spec requires that the sender persona is extracted from the first outbound message in the thread and never switched mid-thread. How would you enforce this across reply, SMS, and follow-up modules.
8. How do you store API keys in ---------- (blueprint variables, Connections vault, other) and why.
9. If you were building this project, what parts would you build first and why.
10. What part of this system is most likely to break if built poorly.
11. How would you structure this so future automations can be added without rebuilding everything.
12. Can you commit to a working demo of Modules 1 to 9 within 48 hours of getting platform access. Yes or no.
13. How fast could you realistically deliver a strong v1.
14. After launch, are you open to a paid monthly retainer plus future builds. What retainer number works for you.
15. Send 2 to 3 examples, screenshots, Looms, or short explanations of similar work.

Write the word BLUEPRINT at the very end of your application.