The project in 30 seconds
- Client
- TechFix
- Industry
- Device & laptop repair
- When
- January 2026
- My role
- Automation engineer
- Channels
- WhatsApp, Instagram, Messenger, web form
- CRM
- GoHighLevel
Where TechFix started
TechFix repairs laptops and phones. Customers reach them however suits them: WhatsApp, Instagram DMs, Facebook Messenger or the website form.
That's good for customers and hard for staff. Someone had to check four inboxes, reply by hand, then re-type the details into the CRM. And once a device was in the shop, customers kept asking the same question: "Is it ready yet?"
What had to be solved
The business needed to:
How I tackled it, step by step
- 1
Connect the channels
Set up Meta developer apps and webhooks for WhatsApp, Instagram and Messenger, plus the website form.
- 2
Give the bot a brain
An OpenAI agent with conversation memory and a structured output parser works out what the customer wants.
- 3
Write to the CRM safely
Search first, create only if needed, for both the contact and the repair opportunity.
- 4
Automate the repair journey
Seven GoHighLevel workflows, one per stage, message the customer and create staff tasks.
Inside the system
15 real screenshots from the build. Click any image to enlarge.
01One AI agent, four channels
Each channel has its own n8n workflow, but they share the same logic. The AI agent understands whether the message is a new repair, a status check, an update or a general question, asks for anything missing, and replies on the same channel.
02Into the CRM, without duplicates
Before creating anything, the workflow searches GoHighLevel for the contact and for an open repair. It only creates what's missing. Every contact is tagged with the channel it came from, so TechFix can see which channels bring in work.
03A repair pipeline that talks to the customer
The Device Repair pipeline has seven stages: lead created, intake complete, diagnosing, quote approval, repair approved, repairing and ready for pickup. Each stage has a workflow that messages the customer and creates the right task for staff.
04Everyone stays informed
Customers get updates in the same conversation thread. Staff get notifications for new tickets and stage changes, so nothing waits on someone checking a screen.
What changed for TechFix
The calls that made it work
Good systems come from good decisions. Here's what I chose, and why.
Structured AI output, not free text
The agent returns a fixed set of fields (intent, device, issue), so the CRM logic never has to guess what the AI meant.
Search before you create
Customers often message twice. Checking for an existing contact and open job first keeps the CRM clean.
One workflow per stage
Small, single-purpose workflows are easier to change than one giant flow when the shop's process evolves.
A similar system could help if…
- Customers contact you on WhatsApp, Instagram or Facebook as well as your website.
- Your team re-types enquiries into a CRM.
- Customers keep asking for status updates.
- You use, or want to use, GoHighLevel.
Customer names, phone numbers and email addresses are blurred. Every screenshot on this page comes from the real project; nothing is a mock-up.