The tasks below have three things in common: they repeat often, they involve text, documents or data, and a person can check the result in seconds. That last point matters. In every example, AI prepares the work and someone on your team decides what happens with it.
1. Sorting and answering incoming emails
A shared inbox like info@ or office@ collects quote requests, complaints, invoices, job applications and spam. AI can read each message, recognise what it is about, route it to the right person and draft a reply in the customer's language.
- Property management: repair reports are recognised, the property and urgency are noted, and a confirmation is drafted for the tenant.
- Fiduciaries: client questions about deadlines or documents are routed to the responsible accountant with a short summary.
What to watch for: start with drafts, not automatic sending. Define clearly which categories exist and who owns each one. Emails often contain personal data, so check where the model runs and that your data is not used for training.
2. Capturing invoices, receipts and delivery notes
Typing supplier invoices into accounting is slow and error prone. AI reads PDFs and scans, extracts supplier, amount, VAT, due date and line items, and prepares the booking for Bexio, Abacus, SAP or your ERP.
- Manufacturing: delivery notes are matched to purchase orders, and differences in quantity are flagged.
- Retail and e-commerce: supplier invoices in several languages are read and prepared for the same chart of accounts.
What to watch for: keep an approval step before anything is booked or paid. Check the error rate on a sample of your real documents in the first weeks. Unusual layouts and handwritten notes still need a human look.
3. Answering customer questions on your website
Many questions come in the evening or at the weekend: opening hours, prices, availability, how to book. A chat assistant built on your own content answers in four languages around the clock and hands over to your team when a question goes beyond what it knows.
- Garages: customers ask about a tyre change or a service slot and receive a booking proposal.
- Hospitality: guests ask about parking, check-in times or allergens on the menu.
What to watch for: the assistant should only answer from content you approve, and say so when it does not know. Plan a clear handover to a person. Tell users they are talking to an AI, and keep the content up to date when prices or hours change.
4. Finding answers in internal documents
Manuals, contracts, internal policies and past projects contain the answers, but finding them takes time. An internal knowledge assistant lets employees ask a question in plain language and get an answer with the source documents shown.
- Insurance brokers: which conditions apply in a specific policy, with the page quoted.
- Manufacturing: maintenance steps from machine manuals, also for new staff.
What to watch for: access rights. Not everyone should see salaries or personnel files, so the assistant must respect the same permissions as your file storage. Answers should always show their sources so people can verify them.
5. Preparing quotes and standard documents
Many quotes follow the same pattern: a request comes in, someone checks the price list, writes the text and adds the conditions. AI can prepare a first version from the request and your templates, which a person then adjusts and sends.
- Trades and building services: a quote for a boiler service, drafted from the customer's email and your price list.
- Logistics: transport offers prepared from route, weight and delivery date.
What to watch for: prices and conditions must come from your system, not from the model's imagination. Keep the templates under your control and always review the final amount before the quote goes out.
6. Translating and adapting texts between national languages
Swiss companies often work in two or three languages. AI translates product descriptions, customer letters, newsletters and internal notes between German, French, Italian and English, and can keep to your own terminology.
- Retail and e-commerce: product texts for the shop in all language versions at once.
- Public administration: notices and information sheets prepared in the official languages.
What to watch for: for legally binding texts, contracts or official decisions, a fluent person should review the translation. Maintain a short glossary of your terms so the wording stays consistent.
7. Writing recurring reports from your data
Weekly sales, open orders, stock levels: the data exists, but someone has to collect it and write a summary. AI can pull the figures from your systems, write a short report in plain language and send it to management every Monday.
- Retail: best sellers, slow movers and items to reorder.
- Property management: open repairs, vacancies and overdue rents per building.
What to watch for: the figures must be calculated by your systems or by clear queries, and the AI only writes the text around them. Check the first reports against your own figures before you rely on them.
The seven tasks at a glance
| Task | Main benefit | Stays with a person |
|---|---|---|
| Sorting and answering emails | Faster replies, nothing lost | Approving and sending the reply |
| Capturing invoices and receipts | No more typing | Approving booking and payment |
| Website questions | Answers also in the evening | Sensitive or unusual requests |
| Internal knowledge search | Answers in seconds | Judging the source |
| Preparing quotes | First version in minutes | Prices and conditions |
| Translation | All languages at once | Binding texts |
| Writing reports | Figures without searching | Interpretation and decisions |
How to start
- Pick one task that repeats often and frustrates your team.
- Collect 20 to 50 real examples, such as emails or invoices.
- Test the solution in a small pilot on real data and measure time saved and errors.
- Decide on the basis of the results whether to take it into production.
This is how our Pilot package works: one use case on real data in about four weeks, from CHF 4,900 excl. VAT. If you are unsure which task to start with, a free 30-minute consultation by video call is often enough to find out, with no commitment.