Skip to content
e-ntegration
Contact

Use cases

Systems from our articles. In each one the agents prepare the decision and a person makes it.

Furniture manufacturing

Five agents, one factory

The problem

A supplier fails, a machine breaks down, workers call in sick. Each disruption sets off a chain of messages between procurement, planning, production, logistics and sales.

What the agents do

Five agents, one per department, read the shared state in ERP and CRM and carry that communication. The agent that owns the problem simulates its options on a copy of the company state, writes a plan naming every change and its cost, and asks its colleagues for their part through A2A.

What stays with people

Each plan goes to the responsible person, who approves it, requests changes or rejects it. The agent then carries out exactly the approved plan.

Read the article
Manufacturing

Quality on the production line

The problem

A coolant temperature alarm sat unread for almost a day. The next morning six parts failed surface inspection.

What the agents do

A line agent answers one question over a rolling seven days: what is hurting our quality this week? It aligns machine state, production orders and inspection findings on one timeline and points to the response your engineers defined.

What stays with people

Operators carry out the response. The agent does not adjust setpoints and does not invent countermeasures.

Read the article
Engineering and research

Engineering decisions over months

The problem

In a long design project, such as a desalination pilot, nobody remembers why a hypothesis was dropped or what depended on it.

What the agents do

The agent is the project's memory: a working wiki and a typed knowledge graph. When a hypothesis is refuted, it reports what falls with it.

What stays with people

The engineer writes the criteria that confirm or refute a hypothesis. The agent stays quiet until asked.

Read the article
Energy infrastructure

From requirements to a binding specification

The problem

A 900-page requirements document for a high-voltage converter station has to be answered clause by clause. One missed requirement loses the bid or becomes a costly obligation.

What the agents do

The agent splits the document into single, numbered requirements, links each one to its source, shows coverage in a compliance matrix and drafts every response from the company's past specifications, in the target language.

What stays with people

The engineer decides each requirement: comply, comply partially, deviate or clarify. There is no button that answers everything.

Read the article
Shipping and long-lived assets

Projection versus reality

The problem

Long-term investments rest on assumptions that go stale silently: fuel prices, regulation, docking windows.

What the agents do

The agent ages every assumption, tracks projection against actuals and brings a review packet before the cheap window to act closes. For irreversible decisions a second agent argues against the leading option.

What stays with people

People judge between the two agents. The agent never re-baselines a plan on its own.

Read the article
Systems engineering

Requirements that argue back

The problem

Domain experts write requirements as prose; engineers build a SysML v2 model. Ambiguity that slips into the model surfaces weeks later as a restructuring.

What the agents do

The agent rewrites prose into the five EARS sentence shapes, compares every draft with everything already specified and only then resolves it against the model. Duplicates and conflicts are found by a reproducible rule it can quote.

What stays with people

When two requirements conflict, the agent quotes both and asks which one is correct. It never chooses.

Read the article
Knowledge transfer

Onboarding that stays current

The problem

Custom training applications only paid off for large intakes and were out of date after the next process change.

What the agents do

An expert teaches the agent directly. It keeps the wiki current, adapts explanations to the trainee, and runs quizzes, role plays and click-through simulations of your systems.

What stays with people

The expert confirms every change to the knowledge base. The agent does not invent a standard or a supplier fact.

Read the article
Industrial distribution

Quotes without the manual lookup

The problem

A quote with about 200 line items took a sales engineer four to six hours across the ERP, a pricing spreadsheet and a supplier database.

What the agents do

Skills for pricing, availability, margin checks and customer history feed an orchestrating agent that drafts the quote.

What stays with people

The sales engineer reviews and approves every quote.

Read the article

Let's talk

Tell us which decision or process your experts need help with.
E-mail
ralph.navasardyan@e-ntegration.de
Address
Weinbergstr. 29, 90607 Rückersdorf, Germany