Use cases
For mid-sized manufacturers of industrial equipment — from Cassel AI, founded by Cord Cassel, himself a manufacturer for thirty years.
Choose the topic where things get stuck in your company. You will then see an example company and the steps in the order in which they build on one another — with one example each. Three example companies, not customer cases: we are at the beginning and say so openly. The examples draw on Cord Cassel's experience. We only analyse emails and conversations with the employees' consent. The works council is involved from the start, nothing is ever measured per person — the goal is growth, not job cuts.
Langdock is the AI platform we set up for your company; from Step 4 your knowledge base is added. What each step includes and costs is set out under Services.
Back to ServicesWhat is the difference? →By topic
Topics without an example for the selected step are greyed out.
What we build for
Knowledge leaves the company when an experienced colleague leaves — and what they know is written down nowhere. It depends on a few people. New staff need months to get up to speed.
In practice: If a named knowledge holder is due to leave, we start with the interviews — knowledge at risk cannot be postponed.
Example company
Test equipment, around 45 people
One applications engineer knows every special solution of recent years. He is due to retire, and what he knows is written down nowhere.
The path in steps
Before
Anyone starting out asks whom to ask. The first months cost the experienced staff more time than the newcomer.
Afterwards you have
The newcomer asks Langdock first — about products, terms, earlier cases — and comes to the colleague with the source, not with an empty question.
What there is time for
The experienced staff are asked about what is written down nowhere. Whatever has been answered before, the verified knowledge answers.
Step 5 · One department in the knowledge base
Before
Anyone can find the standard case in the ERP. Everything that deviates from it still ends up on one person's desk — right to the end. Which order number was which application is written down nowhere.
Afterwards you have
Every order carries a description of the application, and the verified knowledge is searched by it: product line, industry, special feature. The question “Have we had something like this before, and what came of it?” is answered by the verified knowledge — with the source.
What there is time for
The special case turns from archive into memory. Anyone new in sales finds what until now only one experienced colleague knew.
Before
Onboarding documents are out of date as soon as they are finished. New service technicians need months before they visit customers on their own.
Afterwards you have
Onboarding paths and training materials are created from the verified knowledge and stay current when it changes. From day one, the newcomer works with the same knowledge as the experienced staff.
What there is time for
New people are productive sooner, and the experienced staff don't have to explain the same things every time.
Before
A named knowledge holder is due to leave. What they know is written down nowhere — and whatever no one asks about in their last months leaves with them.
Afterwards you have
Structured interviews based on real cases from the company. Each case becomes an entry in the verified knowledge, linked to the order that triggered it. The expert checks what has been made of the interview before it counts. Recordings are only made with consent.
What there is time for
The expert’s experience is recorded and verified: colleagues ask him about what is genuinely new, and whoever comes after him builds on it.
What we build for
Complex quotations tie up the one engineer who really knows the product. No one fully trusts the item master data. Similar orders are designed from scratch every time — and orders are lost because the answer takes too long.
In practice: An obvious place to start. Usually a clean-up of the item master data comes first — no quotation can be built on unreliable master data.
Example company
Special-purpose machinery, around 150 people
Enquiries come in faster than quotations get written. The quotations are written by three experienced engineers — the same ones who are supposed to design new products.
The path in steps
Before
Old quotations and descriptions sit in project folders. Anyone looking for a similar project asks whoever remembers.
Afterwards you have
The 300 most important quotations and descriptions have been reviewed and put in order, together with a list of where they contradict one another. Applications engineering asks: “Have we ever built a palletiser for bagged goods?” — and gets the three projects with their sources.
What there is time for
Searching takes minutes instead of an afternoon. The quotation is still written by hand — and the report you receive from us after four weeks of use shows how much time that costs.
Agent: your people instruct it in the chat. Workflow: starts by itself when something comes in. Both deliver a draft; your people approve it.
Before
Every enquiry arrives in sales as an email. Someone reads it, looks up the similar project and forwards it to the right engineer — often not until the next day.
Afterwards you have
A workflow reads every incoming enquiry, recognises the product line and application, adds the three most similar earlier projects from the verified knowledge and passes everything to the responsible engineer as a draft. What gets forwarded is the engineer's decision.
What there is time for
The engineer starts with the comparison instead of the search. The quotation is still written by hand — an application takes that over in Step 6.
Before
The sales partner gets back a parts list with a price underneath. Why exactly this configuration suits the customer's line is written down nowhere — and many enquiries come to nothing.
Afterwards you have
An agent drafts, from the comparable installations in the verified knowledge, a document explaining what the quotation solves for the customer. The engineer checks it and approves it — on the same day.
What there is time for
The sales partner has an argument in hand, not just a price. The engineer no longer writes the explanation, but checks it.
Before
Every quotation starts with a blank page: looking up items in the ERP, pulling together texts from old quotations.
Afterwards you have
The enquiry becomes a draft quotation — structured on comparable projects, prices and bills of materials from the ERP, every statement with its source. The engineer checks and approves.
What there is time for
Quotations go out while the customer is still deciding. The weeks of your best engineers belong to design again — to the variant two customers have asked for.
Before
The ERP says which configuration was built back then. What happened at the customer afterwards — faults, rework, what they needed later — sits in service emails, complaints and people's heads. At the next similar quotation, no one knows.
Afterwards you have
The draft quotation comes with the history of the comparable installations, from service reports and complaints, every statement with its source. At the quotation stage, the engineer sees what the installation cost later on.
What there is time for
Effort that used to be regularly underestimated shows up at the quotation stage — not at commissioning. Orders you would do better not to win can be recognised while there is still time to decide.
What we build for
Service enquiries always end up with the same few experts — even for installations from twenty years ago. Knowledge across several product generations grows faster than a team can maintain it.
In practice: Works in two places: repeat visits become rarer, and your experts are free for the cases that are genuinely new.
Example company
Packaging machinery, around 140 people
Several machine generations in the field, service questions about all of them. Two colleagues know them; when they are out, engineering gets asked.
The path in steps
Before
The answers to the same recurring customer problems are in the technician's sent emails. Only he can find them — and he is out on the road.
Afterwards you have
The solved cases of one area have been reviewed and can be queried in Langdock. The colleague at the service desk asks about the fault the customer describes and gets the cases in which it was solved — with the source. Only emails that colleagues release for this purpose are analysed.
What there is time for
The experienced technician has his time for the cases that are genuinely new. The service desk answers the standard question itself, while the customer is still on the phone.
Agent: your people instruct it in the chat. Workflow: starts by itself when something comes in. Both deliver a draft; your people approve it.
Before
A support case has four steps: identify the device, look up logs and history, narrow down the problem, write the answer so the customer can follow it. Only the third needs the expert. The other three eat up the hour that pushes the email down the inbox.
Afterwards you have
A workflow assembles the history and the logs for the report. The technician narrows down the problem and sketches the solution; an agent turns the sketch into step-by-step instructions that even a new colleague can follow. The technician approves.
What there is time for
The workflow takes over the hour of preparation. The expert works on the step only they can do — and the email doesn't sit unanswered.
Step 5 · One department in the knowledge base
Before
When a customer calls with a fault, the question lands with the two service colleagues who know every machine. If they are out, the customer waits — or engineering gets asked.
Afterwards you have
For the machine number, the service team has everything to hand: the as-delivered configuration from the ERP, earlier faults and how they were solved from the service reports, the right instructions — with sources. The colleague on the phone can answer the standard question; the experts remain free for the new cases, and every solved case becomes part of the verified knowledge.
What there is time for
The two experienced colleagues are no longer the bottleneck for every fault, and new colleagues in service find their feet faster. With this capacity, you can offer service contracts that no one can staff today.
What we build for
Mistakes from earlier generations repeat themselves. What was once tested, justified or rejected lies scattered across folders, emails and people's heads — and no one finds it again.
Even just by searching your verified knowledge, your designers can ask what has already been built, tested or rejected.
In practice: Suitable as the only starting point only with very clear findings — in most companies, the shoe pinches somewhere else first.
The path in steps
Before
For a new product series, engineering gathers up the test reports of its predecessors. Why a variant was rejected back then is known only to those who were there.
Afterwards you have
A new project is worked through step by step — problem, options, assessment, decision — checked against what the company has already experienced, including rejected ideas. Every decision remains traceable with its reasoning.
What there is time for
Engineering starts with what the company already knows, instead of finding it out all over again.
What we build for
Warranty cases and returns are handled and filed one by one. No one notices that the same pattern runs through a whole product line.
In practice: Pays off once enough verified cases are in the knowledge — so rarely the first step.
The path in steps
Agent: your people instruct it in the chat. Workflow: starts by itself when something comes in. Both deliver a draft; your people approve it.
Before
A known pitfall is recorded in an application note. After the third case, the company knows it. Whether it is taken into account in the fourth quotation depends on who handles it. Known is not built in.
Afterwards you have
An agent checks every enquiry questionnaire against the company's application notes and flags where a customer's information matches a known case — with the relevant note.
What there is time for
Experience is part of every order, not just of the colleague who happens to remember. The follow-up question to the customer comes before the quotation, not after commissioning.
Before
Every warranty case is handled, invoiced and filed. That the third case this year concerns the same component is noticed only by whoever happens to have had all three on their desk.
Afterwards you have
Complaints, warranty cases and service calls are analysed across time, products and suppliers. Recurring patterns become visible, and a quality report is ready as a draft. What the system recognises as a pattern is confirmed by an expert before it counts as fact.
What there is time for
The fault in the third machine is no longer built into the fourth. Quality management sees the line, not just the individual case.
What we build for
Hundreds of real proof points — solved applications, industries served — sit in your systems. To the outside world, your expertise remains invisible.
In practice: Rarely the trigger — but the area where a result becomes visible to the outside world fastest.
The path in steps
Before
For the trade fair, sales needs three application examples. The solved cases sit in project folders and service emails; no one has time to write them up.
Afterwards you have
Application examples, data sheets and statements of benefit are created from what the company has actually solved — and only go out after approval.
What there is time for
Sales has material that comes from your own company, not from a competitor's brochure.
What we build for
More volume needs more of the people you can't simply double — and can't find. Your company can only grow as fast as its few knowledge holders can carry it.
In practice: Not a problem of its own, but what the other six cause as soon as you want to grow.
The path in steps
Before
For a decision on a new product series or a market, someone spends days pulling together figures and experience from different departments — or the decision is made on gut feeling.
Afterwards you have
Scenarios, comparisons of options and key figures from the company's knowledge, prepared for the decisions that rest with management — with the origin of every statement.
What there is time for
The decision stays with you. The groundwork for it is ready in hours, not weeks.
Add-on services
Structured interviews with an experienced expert, based on real cases from your company. The expert checks what has been made of the interview before it counts; recordings are only made with consent. If the expert has already left, we secure what they left behind — their quotations, project files and cases. What was only in their head cannot be brought back. The detailed example is in the topic “Knowledge risk and succession”.
How it starts
Where work piles up, where knowledge depends on individual people — and what that costs you today, calculated with your figures.
Arrange a conversationOr write to Cord Cassel directly: cord.cassel@cassel-ai.com