Book Demo →

KaizenFlow AI Certified Operator — Course

Module 01 · 20 min · 2 fig.

Your shift, on one screen

Your line view is the one screen that tells you how your shift is really going — right now, by station. Learn to read it at a glance and you'll catch trouble while it's still small, before it eats your numbers.

One screen, one station at a time

The live line view lays out every station on your line as a row of tiles. Each station shows three things: its OEE, its downtime, and its scrap. You don't have to add anything up or do math — the screen does that. Your job is to glance across the row and spot which station is dragging.

When a tile turns red or amber, that's the screen pointing at where your shift is losing time or parts. Green means that station is running clean. Start by scanning for the color that doesn't belong.

  • OEE: how well a station is actually running versus its best
  • Downtime: time the station was stopped by a breakdown or fault — changeovers are tracked in their own bucket
  • Scrap: parts made that didn’t pass
FIG. 1.A THE LIVE LINE VIEW KF·CO-01
Filler91%OEECapper63%22 MIN DOWNLabeler88%OEECase packer90%OEE
Four stations, one glance. Three tiles run clean; the capper's red tile and its 22-minute downtime note tell you exactly where to walk.

What each tile is telling you

OEE rolls speed, stops, and scrap into one health number for that station — a low OEE is the headline, not the cause. To find the cause, look at the downtime and scrap tiles next to it.

If downtime is high, the station keeps stopping. If scrap is high, it's running but making bad parts. Read them together: the low tile tells you where to look, the downtime and scrap tiles tell you why.

FIG. 1.B HOW A SHIFT BECOMES AN OEE NUMBER KF·CO-01
100%PLANNED TIME−11%BREAKDOWNS−9%CHANGEOVERS−8%MICRO-STOPS−6%SLOW CYCLES−5%SCRAP + REWORK61%OEE
Every loss bucket takes its cut of planned time. The tile shows you the 61 — the buckets next to it show you which cut to chase.
Worked example — reading one station

The capper ran a 480-minute shift. It was down 72 minutes (availability 85%), ran at 92% of its rated speed while up (performance), and 3 of every 100 parts failed inspection (quality 97%).

OEE = Availability × Performance × Quality = 0.85 × 0.92 × 0.97 = 0.76 → 76%

That 76% isn't a grade — it's a map. The lowest of the three factors (availability, 85%) names the biggest bucket: stops. That's why the downtime tile sits next to the OEE tile.

Key takeaway

Scan for the worst tile first, then read the downtime and scrap next to it — the color tells you where, the neighbors tell you why.

Module quiz · question 1 of 3

Q1One station's OEE tile is red while the others are green. On that station, downtime is high but scrap looks normal. What do you do first?

Module quiz · question 2 of 3

Filler 91% · Capper 63% (22 min down) · Labeler 88% · Case packer 90%.

Q2You walk up to this line view mid-shift. Which station do you walk to, and why?

Module quiz · question 3 of 3

Q3A station's OEE is low, but you haven't looked at any other tile yet. What does that low OEE by itself tell you?

Module 02 · 18 min · 2 fig.

Logging a stop the right way

Every stop, changeover, and scrap part you log becomes the data your plant uses to decide what to fix first. Sloppy logs send the AI chasing the wrong problem; clean logs put the real fix at the top of the list.

Pick the reason that's actually true

When the line goes down, KaizenFlow asks you for a reason code. Don't grab the first one in the list or default to "Other" to clear the screen. The reason code is what groups stops together so the plant can see the pattern behind them.

  • A jam is a jam, not "minor stop" — pick the specific code.
  • If the real cause isn’t in the list, tell your lead so it gets added.
  • "Other" hides a problem; a real code surfaces it.

Time it like it happened

Log the stop close to when it occurs, not all at once at end of shift. Memory blurs, and a stop you log an hour late can land in the wrong bucket. Mark when the line actually went down and when it ran again — that span is the downtime the system measures.

  • Start the log when the line stops, close it when it runs.
  • Changeovers count too: log the start and the finish.
  • Note scrap with its reason while the part is still in front of you.
FIG. 2.A ONE SHIFT, LOGGED CLEAN KF·CO-02
0m60m120m180m240m300m360m420m480mJAMC/OSCRAPRUNNINGDOWNTIMECHANGEOVERSCRAP RUN
A jam, a changeover, and a scrap run — each in its own bucket, logged when it happened. This is what the plant's numbers are built from.

Separate the buckets

Downtime, changeover, and scrap are three different things, and the product tracks them apart. A changeover isn't a breakdown, and a scrapped part isn't downtime. Putting each in its own bucket keeps the Six Big Losses honest, so the next-best action you see is aimed at the loss that's really costing you.

FIG. 2.B WHAT ONE CLEAN ENTRY FEEDS KF·CO-02
Jam — infeedYOUR ENTRY01Groups with 14like itPATTERN02Priced at $/weekAI RANKING03Fix reaches thetopQUEUE04
A specific, on-time reason code groups with its pattern, the pattern gets a dollar figure, and the fix earns a place at the top of the queue. "Other" breaks this chain at the first link.
Key takeaway

Log it now, log the real reason — a stop is only useful data if it's true and on time.

Module quiz · question 1 of 3

The capper jammed at the infeed and stopped the line for six minutes. You cleared it and it's running again. The log asks for a reason.

Q1Which entry keeps the data clean?

Module quiz · question 2 of 3

Q2Why do downtime, changeovers, and scrap go in three separate buckets instead of one "lost time" number?

Module quiz · question 3 of 3

Q3You're slammed and tempted to log the shift's three stops from memory at the end. What's actually at stake?

Module 03 · 22 min · 2 fig.

Acting on the next-best action

Every shift, KaizenFlow looks at your line's OEE, downtime, and scrap and ranks the one fix that will buy back the most time or stop the most loss right now. This module shows you how to read that ranked action and close it out without leaving your station.

How the AI picks the next-best action

The next-best action card sits on your line view. It points at one station and one problem — the thing costing you the most time or scrap this shift, not a list of everything that's a little off. It ranks by impact, so the top card is where your effort pays back fastest.

Each card tells you the station, the loss it's tied to (a downtime reason, a slow run, a changeover, scrap), and what to check. It's a starting point, not an order — you know the line. One rule the ranking can't see, though: a line that's down right now beats any card. If a station stops while you're working a card, clear the live stop first, then come back to the ranked action.

  • Top card = biggest payback right now, not the oldest problem
  • It names a station and a loss type, so you know where to walk
  • A live line-down stop always comes before the card
FIG. 3.A THE RANKED ACTION QUEUE KF·CO-03
01Micro-stops — capper infeed$2,600/wkRECURRING · THIS SHIFT02Changeover overrun — line 2$1,900/wkRECURRING03Slow cycle — case packer$700/wkMONITOR
The top card is the biggest payback right now — not the oldest problem. One rule outranks the queue: a line that is down right now.

Closing the loop on the floor

Acting isn't done until you log what happened. Go to the station, do the check or fix the card suggests, then mark the action on the card — fixed, still open, or not the real cause. Add the reason code if you logged downtime or a changeover while you were there.

That log is what teaches the ranking. When you close an action, the AI sees what worked and re-ranks the next one. Skip the logging and the same card keeps coming back, and the floor data drifts from what really happened.

FIG. 3.B CLOSING THE LOOP ON THE FLOOR KF·CO-03
Read the top cardSTATION + LOSS01Walk & checkYOU KNOW THE LINE02Fix or flagYOUR CALL03Log the outcomeTEACHES THE AI04CLOSED LOOP
The log is what teaches the ranking. Skip it and the same card keeps coming back while the data drifts from what really happened.
Key takeaway

Clear a live stop first, then work the top card — and log what you found, because an action isn't closed until it's logged.

Module quiz · question 1 of 3

You're working the top next-best-action card at the capper. Station 2 jams and stops the line.

Q1What do you do?

Module quiz · question 2 of 3

Q2How does the AI decide which action sits at the top of your queue?

Module quiz · question 3 of 3

Q3You did the check the card suggested, and the suggested cause wasn't it. What's the right close-out?

Module 04 · 20 min · 2 fig.

Catching drift before it’s scrap

A drift is your line slowly sliding out of its normal range before anything breaks. KaizenFlow's anomaly alerts catch that slide early, so you can nudge the process back instead of running a bin of scrap and a quality hold.

What an anomaly alert is telling you

An alert is not the same as a hard fault. It means a reading on your station is moving away from where it normally sits for this part and this shift — trending, not yet failed. The tile shows what drifted (a measurement, a cycle, a temperature, a reject rate), which direction it's heading, and when it started. Think of it as a tap on the shoulder, not an alarm bell.

  • Trend, not failure: the part is still good, but it’s heading the wrong way
  • Read the direction: is the reading climbing or falling, and how fast
  • Note the start time: a drift that began at changeover points you to the cause
FIG. 4.A DRIFT, CAUGHT IN THE WINDOW KF·CO-04
NORMAL RANGEREJECT LIMITALERT FIRESCHANGEOVERCYCLEREADING
The reading leaves its normal band well before the reject limit. That gap — alert to limit — is your window to fix it while it's still a tweak.

Respond while it’s cheap

The whole point of catching drift is that fixing it now is small and fixing it later is a scrap pile. When an alert fires, check the station against the alert first — confirm what KaizenFlow saw with your own eyes or a quick gauge check. If it's real, make the small correction the process calls for and log what you did. If it's a false read or a known one-off, acknowledge it so the next operator isn't chasing a ghost.

  • Confirm at the station before you adjust anything
  • Make the smallest correction that brings it back in range
  • Log the action — a logged drift catch is shift-handover gold
FIG. 4.B THE PRICE OF WAITING KF·CO-04
WAIT FOR A REJECTScrap bin+ quality hold + "when did itstart?"RESPOND AT THE ALERTOne tweakconfirmed at the station · loggedSame driftTWO VERY DIFFERENT SHIFTS
Respond at the alert and the fix is a small adjustment. Wait for a reject and you've bought scrap, a quality hold, and a hunt for when it started.
Key takeaway

An alert means it's drifting, not broken — go look now, while the fix is a tweak and not a scrap bin.

Module quiz · question 1 of 3

Mid-shift, an anomaly tile shows a measurement slowly climbing toward the edge of its normal range. Parts are still passing.

Q1What do you do?

Module quiz · question 2 of 3

Q2What's the difference between an anomaly alert and a fault alarm?

Module quiz · question 3 of 3

Q3You confirm at the station that the alert was a known one-off — a sensor blip after a changeover. What now?

Module 05 · 15 min · 2 fig.

A handover that sticks

The next crew inherits whatever you leave behind. A handover that sticks turns the last five minutes of your shift into a head start for theirs, so the line keeps running instead of re-discovering the problems you already found.

Write the note for someone, not for the record

In the handover screen you're writing to two real people: the operator taking over your line and your supervisor. Both need to act, not just read. So lead with what's still open. If a station is limping or a machine is one stoppage away from going down, say that first, before the routine stuff.

KaizenFlow already carries your shift's OEE, downtime, and scrap forward, so don't retype numbers it already shows. Your job is the context the data can't see: what you did, what you tried, and what's still hanging.

  • Still open: what isn’t fixed and needs eyes now
  • Watch this: a station or machine trending the wrong way
  • Done or changed: anything you swapped, adjusted, or worked around
FIG. 5.A ANATOMY OF A HANDOVER THAT STICKS KF·CO-05
Your handover noteStill openneeds eyes nowunfixed + whyopen AI actionWatch thistrending wrongfix may not holdDone / changedwhat you swappedworkarounds live
Lead with what's open, flag what's trending, note what you changed. The numbers already carry forward — your note carries what the data can't see.

Tie loose ends back to the log

If you logged a downtime, changeover, or scrap event with a reason code during the shift, point the next crew to it instead of re-explaining. "Station 3 jam, see the 2:10 downtime entry, ran fine after the guide reset" is something they can act on in seconds.

If the AI flagged a next-best action you didn't get to, name it and say why: out of time, waiting on parts, needs a supervisor call. That keeps it from getting dropped or repeated by the next crew.

FIG. 5.B POINT AT THE LOG, DON'T RETYPE IT KF·CO-05
The eventSTATION 3 JAM01Your logged entry2:10 · REASON · FIX02Next crew actsIN SECONDS03
"Station 3 jam — see the 2:10 downtime entry — ran fine after the guide reset." Ten seconds to write, ten seconds to act on.
Key takeaway

If the next crew would have to call you to understand it, it's not done — write the open items first, in plain words they can act on.

Module quiz · question 1 of 3

End of shift. Station 3 jammed twice — you reset the guide, but you're not sure the fix will hold. An AI changeover action is still open. OEE, downtime, and scrap already carry forward automatically.

Q1What goes in your handover?

Module quiz · question 2 of 3

Q2Who are you actually writing the handover for?

Module quiz · question 3 of 3

Q3What's the test for whether your handover note is done?

Course complete.

You’ve worked all 5 modules. Sit the free practice exam to see if you’re ready for the real assessment — same format, instant grading, keyed back to the modules.

Take the practice exam → Get certified