Not legal advice. This page sorts public sources, as of 3 Nov 2026. Figures and quotes come from the linked sources. Forecasts are statements by the people named, not facts.
In short
- A building-services firm (heating, ventilation, air conditioning, plumbing) copies prices by hand from Excel into the tender's bill of quantities, a PDF, for every quote. According to the case that takes 4 to 8 hours per quote.
- An estimated 200 working days of savings potential a year. That is a worked example before review time, not a measured result.
- Status: use case discovery, moving to the sandbox. The next milestone is a pilot with the first real client PDF.
- The firm stays anonymous. We name the industry and the workflow, not the company.
Case · as of 3 Nov 2026
The three numbers behind the case
All three are assumptions or estimates from the case, not measurements.
hestimate
4 to 8 hours per quote
- Manual work just to transfer prices into the PDF.
- The calculation itself is not the bottleneck.
#assumption
about 180 quotes a year
- Public and private tenders.
- That is 720 to 1,440 hours a year.
WDworked example
about 200 working days
- 200 quotes × 8 hours ÷ 8 hours a day.
- Before review time by specialists.
As of 3 Nov 2026. Estimate, not a measured result.
The starting point
The bottleneck isn't the calculation, it's the manual transfer. The bill of quantities arrives as a PDF with empty price columns. The prices sit in an Excel calculation. On top come standard texts that repeat: company details, insurance, compliance statements. Today someone fills it in by hand, page by page.
How we get to "about 200 working days"
| Step | Assumption | Result |
|---|---|---|
| Quotes per year | about 180 (assumption from the case) | 180 |
| Manual work per quote | 4 to 8 hours (estimate) | 720 to 1,440 hours a year |
| In working days | 8 hours a day | 90 to 180 working days |
| Worked example | 200 quotes × 8 hours ÷ 8 hours a day | about 200 working days |
Important: this is an estimate, not a measured result. The numbers apply before review time, because the check by specialists stays. If you write more quotes or need more time per quote you end up higher, with fewer you end up lower.
Our stories card puts it the same way (translated): "Worked example from a building-services case: 180 quotes × 4–8 h = 90–180 working days, e.g. 200 working days. Estimate before review time, explained transparently."
What the AI workflow does
- Read the bill of quantities in a structured way.
- Match each item to the right price from the calculation.
- Write prices automatically into the correct PDF lines, with standard texts as template blocks per company.
- Batch run with check log and diff report: item-by-item comparison between calculation and completed PDF, with page and item references.
- A person checks and approves. No quote goes out without approval.
Built so far: an automation spec and test data sets. All details, assumptions and sources are in the research case on quote automation.
What other SMEs can take from it
- Count how often you retype data from one system into another. That's often where the biggest potential sits.
- Calculate transparently: volume × time per task, then subtract review time.
- Start small, with real documents. A pilot shows whether the estimate holds.
More examples: AI agents in SMEs.
From practice: go-lives and cases
What we build ourselves with AI. Client names only where the client published them; cases stay anonymised.
- Eyecatcher: how AI could save e.g. 200 working days on quotes (in German)
- Research case: quote automation for building-services tenders (anonymised)
- Research case: AI quantity take-off for precast concrete
- Compile 26 in London: AI teammates, agents and a year in code (in German)
- AI pilots & case studies from the AI Lab (anonymised)
digihub · In der Specki 3, Schaan
AI in your own business: where do we start?
We don't start with a tool, we start with a bottleneck: which workflow eats noticeable time today? Then we build a pilot with real examples from your daily work and bring it into operation (The Automation Layer). The first conversation takes 30 minutes and is non-binding.
Prefer to work on site? At digihub in Schaan there's coworking with a Flex Desk from CHF 169, a Full Desk from CHF 299 or a Private Office from CHF 549 a month, plus meeting room and community.
Read next
Sources
Not legal advice. Whether and how a rule applies to your business is a question for a qualified adviser. As of 3 Nov 2026.
Content from digihub.li · reviewed 3 November 2026.