Today
Every enquiry for a water treatment plant starts with an analysis: iron, manganese, hardness, turbidity, conductivity, bacteria, usually as a lab report in a PDF, plus the flow per hour and the target, such as drinking water, process water or boiler feed water. Only two or three process engineers can derive and size the right chain of pre-treatment, filter, membrane, disinfection and dosing from it, and they are always booked. Enquiries wait for weeks, and whoever sizes in a hurry misses a value that later fouls the membrane.
An example
A dairy in the Sauerland sends Aldevo Wassertechnik the analysis of its well water and needs 50 m³/h of drinking-water quality for production and cleaning, plus boiler feed water. Iron 1.2 mg/l, manganese 0.3 mg/l, hardness 21 °dH, turbidity 4 NTU, coliform bacteria detected. The agent proposes aeration and an iron removal filter, membrane filtration FA DN 150 with 0.1 µm, UV disinfection at 400 J/m² and softening for the boiler line, and shows two reference plants with similar well water. It flags that the manganese content calls for catalytic filter media and that the analysis lacks the TOC value the membrane design needs. The process engineer sends the TOC question, adjusts the filter stage and releases the draft quote.
Your data, wherever it lives
Today the answers to a water analysis sit in many sources: the lab report as a PDF in a mailbox, product data and operating limits in the PLM, design tables in folders, reference plants and prices in the ERP, and readings from running plants in the lab database. demi reads, understands and links them per analysis value and process stage, with your access rights. The AI proposes chain and sizing in hours instead of weeks, and the process engineer checks and decides.
Where the data lives today
- E-mails
- PDFs
- Office files
- Folders & drives
- PLM
- ERP
- Databases
How demi solves it
Three layers in this order: connect the data first, then the rights, then the AI.
- 1 · ConnectData Layer
Data sources in this case: the enquiry in the mailbox with the lab report as PDF or spreadsheet, your PLM with product data and operating limits, your design tables, the ERP with prices and reference plants, and the lab database with readings from running plants. demi replaces nothing and links analysis values, target quality limits, the operating limits of your filters, membranes, UV units and dosing stations, and earlier plants with similar raw water. For every analysis you find the plants already running on similar water.
- 2 · PermitPermission Layer
Design rules and reference plants are visible to sales and process engineering; prices and costing only to sales. A customer's analyses and plant data are seen only by the team working on that enquiry. The quote goes out only after the responsible process engineer releases the design; the release is recorded.
- 3 · ActIntelligence Layer
The agent reads the analysis, scans included, compares every value with the target quality and the operating limits of your products and proposes the treatment chain. Filter areas, membrane modules, UV dose and dosing rates are calculated by your design rules, not by the language model. If a value the design needs is missing, it writes the question for the customer instead of estimating it.
How it runs
- 1
The enquiry arrives by e-mail with water analysis, flow and target quality.
- 2
The agent checks every analysis value against target quality and operating limits and flags missing values.
- 3
It proposes the treatment chain from your range and shows reference plants with similar raw water.
- 4
Your design rules size filters, membranes, UV and dosing, and the draft quote follows.
- 5
The process engineer checks chain and sizing and releases the quote.
What changes
Each enquiry gets its design in days instead of weeks
Critical analysis values surface before the quote
Experience from running plants goes into every new design
What does it bring you?
Check it with your own numbers
Every starting value is an assumption you can change. The calculation sits right below, step by step.
Your numbers
Result
+216 hfreed per year
Assumptions and calculation
demi proposes chain and sizing from the analysis; your process engineer checks and releases.
4 h → 2.8 h per case
How we calculate
15 design enquiries × 4 h × 30% × 1.0 (manual) × 12 months = 216 h
Value per year, before demi costs
Freed per year+216 h
Assumption: labour cost of €49.50 per hour in German manufacturing in 2025 (Destatis), plus workplace, IT and overheads.
= Value per year, before demi costs
+€14,040
gained per year
216 h ÷ 1,760 h per role = 0.1 full-time roles
An estimate, not a promise. The starting values are assumptions for an industrial company with 150 to 500 people.
Combine with more cases and get the packageQuestions about this use case
Which analysis values does the agent need at minimum?
Your process engineering decides that, per process and target quality. Typical values are turbidity, iron, manganese, hardness, conductivity, pH, organic load and bacterial counts. If a value that a rule needs is missing, it goes into the question for the customer, and the affected stage stays open.
Where do the design rules come from?
From your company: filter velocities, membrane fluxes, UV dose, dosing rates and operating limits per product, as your process engineers maintain them today in spreadsheets and design programs. demi calculates with the released version and shows the rule behind every value. Process engineering adds what it learns from running plants.
Does the agent also check limits and regulations?
Yes, against the target quality you store, such as drinking water regulations, boiler feed water requirements or the customer's specification. Every value above its limit appears with the stage of the chain that lowers it. Whether the chain is enough is decided by your process engineer.
Does this also work for process water and waste water in chemicals?
Yes. Process water, cooling water or pre-treated waste water have other targets and processes, such as reverse osmosis, ion exchange or neutralisation. The agent works with the processes and rules of your range and only proposes what you can deliver.
