Live line visibility in a pet food plant is a real-time view of what every line is doing right now: running or down, at what rate, against what target, and why it stopped. It replaces the end-of-shift paper report with live boards showing rate, downtime, and reason codes so supervisors can act during the shift instead of reading about it the next morning.
Most pet food plants run on hindsight. The line data is real and the operators are diligent, but it lives on clipboards and in spreadsheets until the shift ends, so by the time a supervisor sees that the extruder ran slow for two hours, the two hours are gone. Live line visibility closes that gap. The point is not more data, it is the same data available in the moment a decision can still change the outcome. That distinction, seeing performance as it happens rather than after the fact, is the single biggest lever a plant has over its own day.
What does live line visibility actually show?
Live line visibility shows four things at once for every line: whether it is running or down, the current rate against target, the reason for any stop, and how the shift is tracking toward its production goal. Those four together turn a wall of numbers into a decision. A line that is down tells you to respond; a line running below rate with no stop tells you something is degrading; a shift falling behind target tells you to adjust before the gap is unrecoverable.
In a pet food plant those signals are specific. Rate is kibble throughput or cans per minute against the formula's target. Downtime is a die change, a jam in the coater, a bagger fault, or a scheduled changeover. Reason codes explain each stop in plant language. And the shift target reflects the actual formula running, since a dense therapeutic diet and a light treat do not share a rate. Getting those signals live is the foundation for OEE and for the machine downtime tracking that drives improvement.
Why is a live board better than the morning report?
A live board beats the morning report because it moves the information to where a decision can still change the result. The morning report is an autopsy: it tells you what killed yesterday's output after nothing can be done about it. A live board is a monitor: it tells you the line is bleeding while you can still stop it. Same data, completely different value, because timing is the whole point.
This is exactly the gap the CLS plant closed. Their production data was accurate and thorough, but it lived on paper until the end of a shift, so supervisors could not see performance as it was happening. Harmony made that same data available in real time, and the reporting that used to take manual compilation every morning now generates from the shift data itself. The lesson generalizes to any pet food plant: you rarely need more data, you need the data you already have to arrive in time. That is also the heart of production reporting that works and the visual management discipline behind it.
How do reason codes turn downtime into action?
Reason codes turn downtime from a number into a to-do list by attaching a cause to every stop, so the plant learns not just that it lost time but why and where. A line that was down four hours is a fact; four hours that were sixty percent die changes, thirty percent coater jams, and ten percent bagger faults is a plan. The reason code is what makes downtime actionable instead of merely regrettable.
The failure mode is reason codes that no one fills in, or that all default to a vague "other" bucket because logging them by hand mid-fault is the last thing an operator wants to do. This is where automation earns its place. When the system detects a stop and proposes the likely reason from the machine signal and context, the operator confirms or corrects with one tap instead of typing, and the codes actually get captured. Accurate reason codes then feed the downtime analysis and the andon response that shrink the losses. The connection to agents is direct: an agent that opens the reason code the moment a line stops is the difference between a coded stop and a blank one, which is the theme of AI agents for pet food manufacturing.
What are the steps to stand up live visibility?
Live visibility is less a purchase than a sequence. You connect the signals, define the targets, and build the board people actually watch. Here is the order that works.
- Pick the signals that decide action. Start with run/down status, rate against target, and stop reason. Resist the urge to display everything; a board with fifty numbers is read by no one.
- Connect the sources as they are. Read the extruder and bagger controls, the checkweigher, and the schedule where they live. The goal is to surface existing signals, not to install new instrumentation everywhere.
- Set targets by formula, not by line. A dense diet and a light treat run at different rates. A single line target makes every board wrong half the time. Tie the target to the formula on the schedule.
- Automate reason-code capture. Have the system detect the stop and propose the reason so the operator confirms in one tap. Manual-only reason codes decay into "other."
- Put the board where the crew stands. On the floor, at the supervisor desk, and on a phone. Visibility that requires logging into a report is not live visibility.
- Close the loop each shift. Use the live reason mix in the shift handover so the next crew inherits the real picture, not a clean slate. This is where the board changes behavior.
The order matters because a board built before the targets are right, or before reason capture is automated, becomes a screen people stop trusting. Trust is the whole asset. A board the crew believes changes what they do; a board they doubt is wallpaper.
What do the numbers and standards say?
Live visibility itself is not regulated, but the metrics it surfaces, downtime, rate, and yield, are the same ones behind OEE and the production records your food safety plan relies on. Use these primary references for the metric definitions and the recordkeeping context.
| Reference | What it covers | Source |
|---|---|---|
| 21 CFR 507 records | Records requirements under the animal food rule that production data supports, including monitoring and verification records | eCFR Part 507 Subpart F |
| OEE and Six Big Losses | The availability, performance, and quality framework a live board feeds; downtime and rate are two of its inputs | 21 CFR 507 |
| FDA FSMA animal food rule | Preventive-controls context that makes timely, accurate production monitoring valuable beyond efficiency | FDA FSMA animal food |
The practical point is that the data on a live board is not a separate world from your compliance records. The rate, downtime, and reason codes you watch to run the shift are the same monitoring records that support your food safety plan. Making them live serves both goals at once. For the metric mechanics, see how OEE is calculated.
Where does an operational layer fit?
Live visibility only works if one system can read every source, because the signals that matter live in different boxes: the extruder PLC, the bagger, the checkweigher, the schedule, and the quality log. Stitching those into a single live view by hand is the work most plants never finish, which is why the board stays a someday project.
Harmony AI is an AI-native operational layer built to do exactly that unification. It is agnostic, reading whatever controls and systems you already run without a rip-and-replace, and it surfaces the result as one real-time view. The foundation is in-person and white-glove: Harmony's team stands on your floor, learns which signals actually drive your decisions, and builds the board to fit with AI-assisted agentic coding on a short timeline, rather than shipping a generic dashboard you have to bend your plant around. Agents watch the live signals and, with approval, open the reason code the instant a line stops, so the board is coded automatically instead of relying on manual entry. This is the pattern proven at the plant in the CLS case study, where paper reports became real-time visibility, and it extends across the whole pet food manufacturing operation. To size the cost of the downtime a live board helps you catch, start with the downtime cost calculator, and see the yield side in yield optimization.
Live line visibility is the difference between running your plant and reviewing it. Surface the four signals that drive action, set targets by formula, automate the reason codes, and put the board where the crew stands. Do that and the morning report stops being where you learn what went wrong and becomes a summary of a shift you already steered.