> ## Content Index > Fetch the complete content index at: https://www.thedailyconstraint.com/llms.txt > Use this file to discover other available public pages before exploring further. # How Can Operations Managers Use AI to Find Quick Workarounds When a Process Breaks Down? - URL: https://www.thedailyconstraint.com/how-can-operations-managers-use-ai-to-find-quick-workarounds-when-a-process-breaks-down/ - Published: 2026-08-20T06:04:56.000Z - Updated: 2026-08-20T06:04:56.000Z - Author: Andrea Bullock *This is the fifth post in our six-part series on becoming an irreplaceable operations manager. Last time, we covered breaking down the notes you inherit at the start of a shift and drafting the ones you hand off at the end. This post takes on the fifth item on our list: finding a fast, safe workaround the moment a process breaks down mid-shift.* Line 4 stops. Not on the schedule. Not during changeover. Just stops, mid-run, and every eye in the building finds you inside of ten seconds. You don't get to think about root cause right now. Root cause is a tomorrow problem, or an hour-from-now problem. Right now, you need a fast, safe answer to one question: what do we do until this is fixed? That's the job. Not fixing the machine, you're not maintenance, but deciding what happens to the people, the output, and the orders while someone else fixes it. AI can hand you options fast. It cannot walk onto your floor, see which operator is steady enough to run a manual workaround safely, or know that the temp on line 5 has never touched that machine before. That part is still entirely yours. It has to be. **Give AI the Failure, Not Just the Frustration** ![](https://storage.ghost.io/c/6a/5a/6a5a81c1-c40d-4db0-af99-e0cedc71c0c2/content/images/2026/08/article5_ghost_internal_01_workarounds_before_root_cause.png) You already know the four-part prompt structure from earlier in this series: role, data, constraints, output. For a workaround, constraints do the heavy lifting. AI doesn't know what's non-negotiable on your floor unless you say so, and the fastest option on the list is worthless if it breaks a safety rule to get there. Before you open the chat, get clear on four things: • What stopped, and what's still running around it. • Any hard constraints that can't bend: safety rules, quality specs, lockout and tagout requirements, certifications a task requires. • What you actually have on hand right now: people, backup equipment, manual processes, spare parts. • How much time you have before this becomes a bigger problem: a customer deadline, a downstream line running out of buffer, shift change in twenty minutes. **The Prompt** ![](https://storage.ghost.io/c/6a/5a/6a5a81c1-c40d-4db0-af99-e0cedc71c0c2/content/images/2026/08/article5_ghost_internal_02_give_ai_the_facts.png) Give it the failure, the walls you can't cross, and what's actually within reach. It builds the options from there: *You are an operations management assistant helping me find a short-term workaround for a process failure. This is not a permanent fix, root-cause analysis happens later. I need options I can act on in the next few minutes.* *Here is what happened: \[describe what broke and what's still running around it\].* *Here are my hard constraints: \[safety rules, quality specs, certifications, lockout and tagout requirements that cannot be bent\].* *Here is what I actually have available right now: \[people, backup equipment, manual processes, spare parts\].* *Structure your output using the exact sections below:* *1\. IMMEDIATE OPTIONS* *\- Three or four workaround options, ranked fastest to slowest to put in place, using only what I said I have available.* *2\. SAFETY AND QUALITY CHECK* *\- For each option, flag anything that could compromise safety or quality. If an option requires bending a constraint I listed, say so plainly and rule it out.* *3\. WHO NEEDS TO EXECUTE THIS* *\- For each option, who needs to be involved, and any skill, certification, or access they need that not everyone on shift has.* *4\. WHAT THIS BUYS YOU* *\- For each option, roughly how long it realistically holds before it stops working or creates a new problem.* *5\. WHAT NOT TO DO* *\- Anything that looks like an obvious shortcut but tends to cause a bigger failure later, based on what I described.* *Do not suggest anything that violates a constraint I listed, even if it would be the fastest option.* **You're Still the One Who Decides** ![](https://storage.ghost.io/c/6a/5a/6a5a81c1-c40d-4db0-af99-e0cedc71c0c2/content/images/2026/08/article5_ghost_internal_03_you_make_the_call.png) AI just handed you four ranked options in under a minute. That's fast. It's also incomplete, because it doesn't know your building. Go back to the first post in this series: you know the people on your team, you know the equipment on your floor, you know which operator can be trusted to run something manual safely and which one needs a second set of hands nearby. AI can't see any of that. You still choose. *That's not AI falling short.* That's the whole point. AI can shrink the time it takes you to see your options. It can't shrink the judgment call underneath them, and it shouldn't. Pick the option, and note why you picked it. That reasoning is worth more than the list itself the next time something similar happens. **What Comes Back** Say Line 4's primary conveyor motor failed and there's no spare on-site until tomorrow. Feed that in, along with your constraints (no bypassing the E-stop circuit, product can't sit unrefrigerated past 30 minutes) and what you have on hand (two floaters, the old manual roller table in the back, one operator certified on manual pallet jacks). Here's roughly what comes back. • IMMEDIATE OPTIONS: Route product manually on the old roller table (fastest, no wait). Shift Line 4's volume to Line 5 if it has spare capacity. Hold product at the staging buffer if it's under the 30-minute window. • SAFETY AND QUALITY CHECK: The manual roller table needs two people minimum, one on each end, per your posted safety guideline. Flag headcount before starting. • WHO NEEDS TO EXECUTE THIS: Your certified pallet-jack operator, plus one floater. • WHAT THIS BUYS YOU: Manual routing holds for roughly two hours before fatigue and pace become the next problem. • WHAT NOT TO DO: Don't try to push the motor past its E-stop trip. That's the constraint you already flagged as non-negotiable. Five minutes of frustration standing at a dead conveyor turns into a decision you can actually stand behind, and a clock on how long that decision holds. **If None of the Options Are Good** Sometimes nothing on the list works cleanly. Every option costs too much time, needs a person you don't have on shift, or brushes right up against a constraint you're not willing to bend. That's not a failed prompt. That's real information: this problem is bigger than a shift-level fix. When that happens, stop looking for a workaround and start drafting the message that gets the right person moving faster. You already know how. Post two in this series covered exactly this: one situation, multiple audiences, each getting only what they need. Feed AI the same failure, minus the “find me a workaround” framing, and ask it to draft the escalation instead: what broke, what you've already ruled out and why, and what you need from whomever is above you to make a real decision. Knowing when to stop looking for a workaround and start escalating is itself the judgment call. AI won't tell you which moment that is. You will. **Fine-Tune It Fast** If the first batch of options doesn't fit the shift you're actually running, say so: • “Give me only the options that don't need a second person. I'm running short-staffed.” • “Assume I have 20 minutes before shift change. Which of these do I start now?” • “Rewrite the safety check assuming this is happening on nights, with fewer people on the floor.” • “I ruled out option two. Give me two more that don't touch Line 5.” Every correction you make teaches you a little more about what to put in the prompt up front next time. That's the whole series compounding, not just this one prompt. **Where This Feeds Back into the Loop** The workaround you pick today becomes raw material for the rest of what you're building in this series. It's the item that goes in your end-of-shift note under WATCH THIS or STILL WAITING ON. It's very likely the root of tomorrow's Gemba question too: why did this fail in the first place, and is the workaround quietly becoming the new normal because nobody's chasing the actual cause? A good workaround buys you time. It was never supposed to be the fix. **Next Up in the Series** We're five prompts into the six-part series now. If you're just catching up, the earlier posts covered turning morning notes into a to-do list, getting one clear update out to every stakeholder, sharper Gemba questions, and building a clean shift handoff in both directions. Next up, the final post in the series: translating precise instructions for employees who speak little English. **Try This on Your Next Breakdown** Try this the next time something breaks down mid-shift, before you've had time to think. Don't tell me the options were fast. Instead, *comment below* on whether picking from AI's list got you to a safe call quicker than usual, and whether it ever handed you an option you had to shut down because you knew something about your floor it didn't. Subscribe to The Daily Constraint to learn more about how you can use AI as a tool to get more for your time and effort as an operations manager, making you truly irreplaceable to your organization.