Scrap analysis with AI: where, when and why scrap occurs
Every production site knows its scrap rate. The question of why it is higher in one shift than in another is usually answered by a spreadsheet that someone builds by hand once a month. With Onagero, your production team describes it once, and the tool then recalculates it every time it is opened, from the current postings.
- Runs on your server
- Reads only, never writes
- GDPR ready
- First result in weeks
In short
Onagero links your scrap postings to order, line, shift, tooling and material, and your team builds a scrap analysis from them, with the data as-of date above every view. You see where scrap occurs and which characteristics appear together. Your production team assesses the cause; the tool provides the trail.
How your team describes it
Your experts say what they need in their own words. Onagero finds the right data and builds the tool in a test stage; after your release it runs in operation.
„Show me where the scrap of the last four weeks came from, by line, shift and tooling, and what the bad days have in common.“Analyze scrap, as a tool in the browser
What the tool takes over
Three steps done by hand today. The tool your team builds with Onagero then does them itself.
Merging
Scrap postings, confirmations, routings and material batches sit in different tables. Onagero links them via order and time, defined once and then automatic.
Breaking down
Scrap by line, shift, tooling, item and material, side by side and over weeks. What moves together and what does not is visible, instead of disappearing in the average.
Asking follow-up questions
Why did the rate rise in week 37? The answer shows the postings it is based on and flags what the data cannot answer.
Typical situations
This is what the value looks like day to day.
Night shift stands out
The night shift rate has been higher for weeks. Is it the shift, or are different items running at night? The tool separates the two.
After a tooling change
After a tooling change, scrap rises for a few hours. How long exactly, and for which tooling does it last longer? The confirmations show it.
New material batch
A supplier batch stands out. The tool shows which orders were produced with that batch and what their scrap looked like.
Monthly review
Instead of a spreadsheet that someone spends two hours preparing, production management opens a tool that is already finished and shows its sources.
Which data is needed
- Scrap or defect postings with quantity, reason and time
- Confirmations per order and operation (ERP or BDE)
- Shift model: when each shift starts
- Optional: tooling and material batches per order
Only what your IT releases is read. Names are masked in the development stage. How the data path works
What you end up with
An internal tool in the browser that shows the scrap of recent days by line, shift and item, flags anomalies and opens the underlying postings for every figure. Plus a written definition of what counts as scrap, released by your production team.
Start with one problem.
Keep the capability.
No AI transformation. One real process, solved on your data in a few weeks. Your team solves the next one itself.
How your price is formed
After the demo you get a fixed offer. Scope, goal and price are set in writing before it starts.
One real problem, the data sources it needs, your Onagero environment on your server, and the first solution, built with your team and measured against the agreed goal.
- Scope, goal and price in writing before the start
- First running solution in a few weeks
- Your data and results stay exportable at any time
Operating the platform on your server: monitoring, reviewed updates, backup and rehearsed recovery, and maintenance of the connections when systems change.
- Begins only once the start has reached its goal
- Your team builds further solutions itself, in the same environment
- Roles, releases and logs included
What you can rely on
The tool reads, computes and shows. These things it never does, and that is in the contract.
- Onagero does not intervene in machine controllers or process parameters.
- The tool does not assess individual employees. Evaluations by shift are agreed with the works council before they run.
- A cause is a finding of your production team, not a result of the model. The tool shows relationships, not blame.
Questions about Analyze scrap
How is this different from an Excel evaluation?
The Excel evaluation is built by hand and is out of date the next day. The tool fetches the data itself, calculates by fixed rules and stays current. You can still export at any time.
Does the AI identify the cause itself?
No. It shows characteristics that occur together with elevated scrap, such as a tooling or a batch. Whether that is the cause is checked by your production team. Occurring together is not proof.
How precisely must the scrap reason be posted?
The more precise, the better, but a reason code is enough to begin with. The tool can additionally organize free texts; your quality assurance checks the classification.
Can we count start-up parts separately?
Yes, if they can be distinguished in the data, for example by a reason of their own or by the time gap since setup. We define this rule together with you.
Does the tool show employee names?
Not in the development stage; names are masked there. Whether the finished tool shows people is something you decide together with the works council. For scrap analysis this is generally not necessary.
How current are the figures?
As current as your source. An update per night or per shift is usual. The data as-of date is shown above every view.
What does a scrap analysis cost?
After the demo you receive a fixed offer, tailored to your lines and data sources. You decide on ongoing operation only once the agreed goal has been measured.
Do we need sensor data for this?
No. The analysis is based on postings and confirmations. Sensor data can be added later if it sits in a database that your IT releases.
A direct line
Write two sentences about Analyze scrap. You get an answer from a person who builds the tool themselves.
Prefer something structured? The fit check asks seven questions and tells you right away whether a pilot fits.
Which problem would
your team solve first?
Show us a real process. We will show you how it could look with Onagero.
Book a demo →