If you're an Apple technician, you've had this moment: a customer drops off a device with a vague description, you've got eight other jobs on the bench, and you need to triage fast. The old workflow is to rely on experience, search a forum, or ask a colleague. It works — but it's slow.
AI changes this. Not by replacing your diagnosis, but by giving you a structured starting point in seconds. Here's exactly how to use it.
The Problem With Traditional Triage
Diagnosis is the most cognitively expensive part of the job. Every device is a context switch — different model, different symptoms, different history. For experienced technicians this becomes intuition over time. For junior technicians, it's guesswork until the pattern recognition builds up.
The bottleneck isn't knowledge. It's time. Searching forums, waiting for a colleague to be free, cross-referencing symptoms manually — all of it adds up across a busy day.
AI doesn't replace the experience. It compresses the research phase.
Why AI Works for Diagnostic Triage
Modern AI models have been trained on enormous amounts of technical documentation, repair community discussions and device behaviour patterns. When you describe a symptom clearly, a good AI model can return a ranked list of probable causes, what to check first and what components are typically involved.
It's not a replacement for hands-on diagnosis. It's a second opinion before you open the device.
The key word is clearly. Vague input gets vague output. The quality of what comes back depends almost entirely on how well you describe the problem.
How to Build a Good Diagnostic Prompt
A good diagnostic prompt has four components:
1. Device context — model, generation, OS version. The more specific the better. "iPhone" is useless. "iPhone 14 Pro, iOS 17.4" is useful.
2. Reported symptoms — exactly what the customer described, plus what you observed yourself. Separate the two.
3. What you've already ruled out — this is the part most people skip, and it's the most valuable. Telling the AI what you've already tested stops it from suggesting things you've already done.
4. What you want back — be explicit. Ask for a ranked list of probable causes, what to test first, and what components may be involved.
Here's the structure in practice:
I'm an Apple technician. Device: [model, OS version]. Customer reports: [symptoms]. I've already ruled out: [what you tested]. Give me a ranked list of probable hardware causes, what to test first, and what components may need replacement.
That's it. Simple, repeatable, consistent.
3 Real Examples
Example 1 — iPhone
Prompt: I'm an Apple technician. Device: iPhone 14 Pro, iOS 17.4. Customer reports screen goes black randomly — sometimes during calls, sometimes when idle. No physical damage visible. Battery health 91%. Already ruled out: software issue (restored to factory settings). Give me a ranked list of probable hardware causes, what to test first, and what components may need replacement.
What AI returns: display connector issue, logic board fault at display power circuit, proximity sensor interference. Suggests starting with a known-good display swap to isolate.
Example 2 — Mac
Prompt: I'm an Apple technician. Device: MacBook Air M1 2020, macOS 14.3. Symptom: on wake from sleep, display stays black. Backlight is on — visible with torch. External display works fine via USB-C. Keyboard and trackpad respond. Happens roughly 30% of wake cycles. No physical damage. Already ruled out: software (reinstalled macOS). Give me a ranked list of probable causes and what to investigate first.
What AI returns: display connector seating, lid angle sensor fault, display assembly issue, known firmware interaction on early M1 units. Suggests checking connector before ordering parts.
Example 3 — iPad
Prompt: I'm an Apple technician. Device: iPad Pro 11-inch 3rd gen, iPadOS 17.2. Customer reports touch unresponsive in bottom-left quadrant. No cracked screen, no drop history reported. Already ruled out: software restart, reset settings. Give me probable causes and first steps.
What AI returns: digitiser layer fault in that zone, display assembly delamination, potential connector issue. Recommends testing with known-good display before quoting.
Common Mistakes to Avoid
Being too vague. "iPhone not working" will return generic suggestions. Specificity is everything.
Skipping the ruled-out section. If you don't tell the AI what you've tested, you'll get suggestions for things you've already done.
Treating the output as a diagnosis. AI gives you hypotheses. You verify them. Never order a part based solely on AI output.
Using it for software issues without context. AI is most useful for hardware fault patterns. For software, the Apple diagnostic tools are still your first stop.
The Workflow in Practice
The best way to use this is at intake, before you open the device. Take the customer's description, add your own observation, and run the prompt while the customer is still in front of you. By the time they've signed the repair order, you already have a mental map of where to start.
It won't replace experience. But it will make your triage faster, more consistent, and less dependent on whether you've seen that exact failure before.
Want the Full Prompt Pack?
This article covers the diagnostic prompt. There are four more — for writing repair estimates, handling difficult customer conversations, researching unfamiliar failures and building your own SOPs.
They're all in the free guide: AI on the Bench — 5 ways to use AI in your Apple repair workflow.
BenchNotes publishes practical tools and guides for Apple technicians every week. Subscribe free to get them in your inbox.