Value and effort

How much preparation does your organisation repeat?

When teams search the same sources and clarify the same policies for similar tasks, preparation is repeated. NomOS makes approved knowledge and reviewed decisions reusable. Whether that is worthwhile for your workflows also depends on setup and ongoing maintenance.

How NomOS supports a specific task.

This fictional quote example contrasts a general answer with a proposal based on an internal policy and approval path. The answers and blocked action are simulated. Technical controls depend on the integration; the example does not run a model.

Fictional example What discount may I give this customer on the annual package?

Without an applicable basis

Discounts between 10 and 20 per cent are common, and often more on annual contracts. For a strategically important customer you can aim at the upper end and justify it over the contract term.

The answer sounds plausible but does not account for your internal rule.

With NomOS

Up to 15 per cent you can approve yourself. Above that the division lead has to approve, and the basis is the current price list, not last year’s prices.

Source
Price framework 2026, section 4
Responsible
Division lead sales
Validity
in force since 01/03/2026
Rule applied
Discounts above 15 per cent need approval from the division lead.

Send the quote with a 22 per cent discount

Blocked in this example

Reason: The discount is above the limit you may approve yourself, and no approval from the division lead is on file.

Exception path: Request a time-limited exception from the division lead, with a reason and an expiry date.

Build on reviewed work in the next case.

The impact graph connects the source used with the proposal, approval and decision. A reviewed basis can be reused as long as it still applies. Changes, exceptions and corrections still require subject-matter review.

Include these measures in the comparison.

  • Preparation and handling

    How long do comparable cases take today and in the pilot, including search and follow-up questions?

  • Quality and rework

    Which proposals need correcting? Can the sources and approvals used be traced?

  • Setup and ongoing maintenance

    How much work do IT and process owners spend on integration, source review and policy changes?

  • Cost and reuse

    How do platform, model and operating costs compare with usable time savings? Do not count increased throughput again as a full time saving.

In the 30-day pilot, we compare two real workflows with your current way of working. We agree on evaluation criteria, scope and a fixed price before starting. The joint evaluation shows whether adoption makes sense.

Retain the value of AI work

Agents submit findings through MCP for review in NomOS. Once adopted, they are available for further tasks. What gets reused is the knowledge and its rationale. We measure how much repeated research and model work this avoids, including retrieval, updates and review.

How agents contribute knowledge ↗
Calculate with your own assumptions.

The calculator estimates working time and model context from your inputs. Example values are not customer results. The reusable share describes your assumption, not measured savings.

Illustrative values are prefilled. Replace them with your own assumptions.

Your organisation
Effort per question
Model change
Your assumption, not a NomOS price. Include setup and ongoing maintenance.

Annual effort spent repeatedly rebuilding context

 

 

Halve or double the minutes per question:  

What makes up this amount

  • Working time 
  • Tokens 

Reusable share based on your assumption per year

Cost of a model change one-off

Enter your assumed annual cost above.

Scenario estimate, not measured savings or investment payback. Reused knowledge still requires model context.

Your assumptions and results
ItemValue
How this is calculated

Working time: people × questions per week × 46 working weeks × share needing context × minutes per question × hourly cost. Tokens: the same number of questions × tokens per context ÷ 1 million × price, at roughly 600 tokens per page of prose. Model change: person-days × day rate. A person-day is eight hours.

All results are based on the values visible above. The fields start with illustrative examples, not NomOS benchmarks. The reusable share is an assumption, not a measured saving. The month count divides the entered annual cost by this assumed annual benefit and multiplies by twelve. It is rounded to whole months, at least one month, and is not an investment payback calculation. Reused knowledge still requires model context.

The calculation runs entirely in your browser. No input is transmitted and nothing is stored.