Cut fault diagnosis by 90% on every machine you have delivered.
Your technician scans the cabinet, asks what the fault means, and gets the answer with the PLC line and drawing page it came from. More calls closed on the first visit, fewer vans back on the road, and the diagnosis your best engineer would give available to everyone on the team.
On this delivery the infeed pusher retracts on the rear proximity switch ‑31B2. A‑4021 is raised when that input stays low past the 1.5 s window in FB 812 Infeed, so the pusher is commanded back and never reports home.
Delivery D‑217 was commissioned with the revised bracket (drawing rev C), which sits the switch 4 mm further back than the standard machine.
Every fault your technician cannot place is a second visit you pay for.
The machine is at the customer's plant, three hours away. The alarm text is generic. Your technician carries the manual for the machine family, not for this delivery, and the electrical drawing that matters was revised twice during commissioning. So the repair becomes an educated guess. The line runs again for a shift, then stops the same way on Thursday.
Somebody drives back. The second trip is not on the invoice. It comes out of the margin on a service contract you already priced, and out of a technician who was booked somewhere else that day.
The answer usually existed. It sat in a PLC program your technician did not write, on page 47 of a drawing that lives in the folder for that delivery, or in the head of the colleague who commissioned the machine in 2014 and is on holiday until the 20th.
Your second-year technician diagnoses like your best one.
One project per machine or line, indexed from the documentation you already produce. A QR code on the cabinet opens that machine's chat. Your technician scans it and asks in plain language, in the language they speak.
- 01 Hotline
The call comes in and does not automatically become a trip.
Whoever answers the phone runs the query on that delivery while the customer is still on the line. Some calls end there with a setting changed or a bracket reseated. The ones that do not end there produce a technician who already knows what they are driving to.
Same answer as the technician gets No trip booked yet - 02 Dispatch
The van gets loaded for the machine that is actually out there.
The same machine family goes out twenty times and no two are identical. Each delivery is its own project, so the answer is about the revision that shipped on that one, down to the bracket the commissioning engineer changed on site. Which means the right part goes in the van the first time.
Per delivery, per revision As-built, not as-designed - 03 At the machine
Scan the cabinet, ask in plain language, get the line it came from.
No laptop balanced on a guard rail and no VPN into the customer's network. A code on the cabinet door opens that machine's chat on the phone your technician already carries, and the answer quotes the PLC block, the line number and the drawing page rather than paraphrasing them.
Cited every time In the technician's own language - 04 Closing the visit
The fix stops living in one person's memory.
"We have seen this before on another delivery" becomes a query instead of a recollection. Edmund reads the service records and journals you already keep, cites the visit the fix came from, and surfaces the fault that keeps coming back across the family.
Repeat faults surfaced Handwritten notes included - 05 The whole base
Your senior engineer stops being the escalation path for everything.
Most of what that person knows was written down at some point. It was just unfindable across a decade of delivery folders. Indexed, a technician in their second year answers the same fault, and the week of the one irreplaceable engineer goes back to the deliveries that need them.
Onboarding a new technician Retirement stops being a risk
It reads the documentation you already produce.
Edmund is vendor-agnostic on purpose. You do not build one machine with one controller, and neither do the plants you deliver into.
Reads the program you shipped, whichever vendor is in the cabinet.
TIA Portal, Allen-Bradley and Omron projects parsed into one searchable model. Blocks, tags, addresses, call sites. The answer quotes the line, and your technician can navigate code written by a colleague eight years ago.
The home check runs on a fixed 1.5 s window in FB 812 Infeed. If ‑31B2 has not made by then, the alarm latches and the retract command is dropped:
201//Retract, home check 202IF #Retract AND NOT #Stop THEN 203 #tHome(IN := NOT #In_31B2, 204 PT := T#1500MS); 205 IF #tHome.Q THEN 206 #alarm.A4021 := true; 207 #cmd.retract := false; 208 END_IF; 209END_IF;
Turns a tag into a place your technician can put a hand on.
EPLAN and CAD exports, down to the terminal and the cabinet. Ask where a tag is wired and get the input and module it lands on. The terminal strip and the drawing page come with it, for the revision that shipped on that delivery.
‑31B2 is the pusher home switch (inductive, PNP). It lands on input %I4.3 of DI module ‑A4, through terminals X1:11‑12 in cabinet +SK3.
Cites the page, including the binders nobody has opened since 2009.
Vendor documentation answered with a page-level citation. Pages with no text layer are read with OCR, and the ones that stay unreadable are reported rather than quietly skipped, per delivery.
"Sensing distance is specified for a standard steel target. Non-ferrous targets and mounting brackets that shift the target beyond 4 mm reduce the effective range and can prevent the output from switching at nominal 2 mm Sn."
- Case packer CP-12 · operating manual
- Infeed sensor datasheet
- Bracket drawing rev C (scan)
- Gearbox manual (scan, 2009)
The fix from delivery D-191 answers the call from delivery D-217.
Service history and maintenance journals, structured or handwritten, become part of the same answer. Edmund reads them as a source and cites the visit. It does not replace the system you log them in.
You see the gaps before a technician finds them at the machine.
Every project reports what is indexed and which pages came in as a scan with no text layer behind them. The cleanup you decide to do afterwards is aimed at the deliveries that actually generate callouts.
- Electrical documentation 98%
- PLC and HMI 100%
- Vendor manuals 91%
- Legacy scans, pre-2012 62%
Ask the line what it has been doing since you left.
Connect the databases behind the machine and ask about them in plain language. Cycle counts, consumption, how often that alarm has fired since the last visit, so a service call starts from a trend instead of a phone description.
34 times across 8 weeks, and the rate roughly tripled from week 5. The step lines up with the shift to the heavier carton grade on 14 April.
What repeat visits cost you this year.
Your inputs, our arithmetic. We are not putting a fix rate on a marketing page. Below is what the second visits cost you before anything changes, and what a quarter of them are worth if they stop happening.
Counting one technician-day per visit. Change any number to whatever your dispatch board actually says.
Most builders do not measure their first-visit fix rate yet. Working out what yours is takes about twenty minutes on the call, and it is useful to you whether or not you buy anything.
Four sentences. Any one of them means we should talk.
Tick what is true of your service organisation. If none of it is, you probably run fine without us, and we would rather say that on a call than walk you through a funnel.
26% lower repair time, measured in plants running Edmund.
Edmund started inside factories, with maintenance teams. That is where the figures below were measured and that is how they are labelled. The fault is the same fault. What differs is who stands at the machine and who pays for the trip.
90% faster diagnosis. 440 technician-hours back per plant per year.
Read the caseManuals, maintenance records and live SQL answered from the shop floor.
Read the caseOn the builder side we have customers and no published figures yet, so we are not going to invent any. If you would rather hear it from a machine builder than from us, say so on the call.
The customer's own maintenance people can use it too.
A code on the cabinet works for whoever scans it. If you want the plant's maintenance team answering the easy questions themselves, that is a matter of who you give access to, and access follows teams rather than the machine.
Fewer calls reach your hotline. The ones that do are the calls worth sending a technician to. Your service organisation stops being the first line of support for a loose sensor bracket and stays the escalation your contract actually charges for.
Whether that sits inside your service contract or beside it is a commercial question. We will have it on the call.
One machine family, your own faults, before you commit to anything.
Not the whole installed base. One family you deliver often, and the documentation you already produce for it.
The ones service directors ask in the first ten minutes.
Our documentation is a mess. Every delivery folder looks different.
That is the normal starting state and it is not a reason to wait. We index one machine family as it is, then show you what comes back before you reorganise anything. Edmund reports per project which documents are indexed, which pages carry no text layer, and where coverage is thin, so the cleanup you do afterwards is aimed at the gaps that actually cost you visits.
Half our drawings are scans from before we had EPLAN.
Scanned pages are read with OCR and cited by page like any other document. Pages that stay unreadable are listed as unreadable. You will know which parts of the old base are covered and which are not, per delivery.
Our customers' PLC programs are under NDA. Who sees what?
Each delivery is its own project, and which projects a person can reach is decided by the team they are in. A technician assigned to one customer's team does not see another customer's projects. Administrators define projects and teams, so the separation is yours to set, per customer if that is what the NDA requires.
The contractual side, where the data is held and what may be done with it, is a document we send you rather than a paragraph in a marketing FAQ. Ask for it on the call and you get it before anything is uploaded.
Our diagnostic know-how is the service margin. Does it leak?
Your projects belong to your organisation in Edmund. Another builder's account does not reach them, and your know-how does not become part of anyone else's answers.
The change worth thinking about is internal. Today that know-how sits with two or three people, which feels like protection until one of them retires. Indexed, it stays inside your service organisation and survives the people who built it.
We already run a service ticketing system.
Keep it. Edmund reads service records and maintenance journals as a source and cites them in answers. Your technicians log work where they log it today. What changes is that the last five years of those records become answerable at the machine instead of searchable by whoever remembers the right ticket number.
Our technicians will not use another system.
There is nothing to log in to at the machine. A code on the cabinet opens that machine's chat on the phone already in their pocket, and the interaction is a question in their own language. The two technicians in the pilot are the test of this, which is why we ask for the sceptical ones.
What does it cost?
Priced by the number of machines indexed and the number of people using it. We quote once we know those two numbers, which usually takes one call. If your repeat-visit figure from the calculator above is small, we will tell you the case is thin rather than sell you the pilot anyway.
