Entry points

The ten most common starting points.

Each one can be done on its own, and each can be the small first thing you start with. The effort and payback figures are general estimates that vary by company; the risk column is the part we most want you to read before you begin.

一Quotations and contract drafts

For companies producing ten or more quotations or contract drafts a week: trading, professional services, contracting.

Effort: Turn your three to five best past examples into templates; AI drafts, a person reviews and signs off. One to two weeks.
Payback: Within two weeks. A first draft goes from half an hour to a few minutes.
Risk: Template quality decides output quality; amounts and terms must be checked by hand — never let AI set the price.

二Client emails and follow-ups

For teams answering dozens of similar emails a day, with follow-up rhythms kept in someone's memory.

Effort: Write a dozen reply skeletons for the common situations. One week.
Payback: Immediate. The harder part is the habit: AI drafts first, a person edits.
Risk: Tone drift — clients can spot a template. For important clients, write the last sentence yourself.

三Meeting and client conversation notes, turned into actions

For high-ticket sales (insurance, property, wealth management), consultants, and anyone whose business runs on conversation.

Effort: Transcribe recordings or notes and let AI produce a summary, to-dos and next steps. One week to build the habit.
Payback: From the first meeting. After a month, what clients said is kept for the first time.
Risk: Recording requires notice and consent; personal client data never goes into a tool without consent.

四Internal knowledge base with Q&A

For companies whose key know-how lives in a few senior heads and new hires learn by word of mouth.

Effort: Gather scattered documents, emails and records into one place, with AI search and Q&A on top. Three to six weeks.
Payback: One to two months. Newcomers' common questions get an old hand on call at any hour.
Risk: Unsorted data gives untrustworthy answers; one person has to own ongoing maintenance.

五Automated weekly business report

For companies where the owner sees last month's numbers in the middle of this month.

Effort: Connect POS, inventory and bank records into a one-page weekly report; AI translates the numbers into plain language. Two to four weeks.
Payback: Every week once built; decisions move from half a beat behind to the same week.
Risk: Inconsistent data sources produce professional-looking wrong numbers; every anomaly needs a human asking "why".

六Inquiry and needs analysis

For trading and service companies that rush to quote the moment an inquiry arrives.

Effort: Let AI break down each inquiry first — background, parameters, what's unclear, possible contradictions — and reply with questions instead of a price. One week.
Payback: From the first inquiry. Buyers with a real project answer seriously; the rest filter themselves out.
Risk: The quote itself cannot be delegated: exchange rates, costs and stock go stale — a person checks before it's sent.

七First-pass contract and document review

For companies where the owner reads every contract word by word.

Effort: Build a review checklist (payment terms, liability, deadlines, unusual clauses); AI reviews against it and flags what a person must read. One to two weeks.
Payback: Immediate. The owner goes from reading everything to reading the flags.
Risk: AI misses things and over-flags others; a first pass is not a final pass — a person reads before signing, and major contracts go to a lawyer.

八Cross-border and multilingual communication

For companies with clients and suppliers across different language markets.

Effort: Almost none. The hard part is building a glossary and a voice so translations read as if one of your own wrote them. One week.
Payback: Immediate.
Risk: Contracts and technical specifications need a second check; politeness levels and forms of address are things AI often gets wrong.

九Social content and professional article drafting

For professionals who know they should publish regularly and never find the time.

Effort: Record your thoughts first (voice is fine), let AI shape a draft, then rewrite it in your own words. One to two hours per piece.
Payback: From zero to one piece a week within a month.
Risk: The AI flavour — readers can tell. Publish it untouched and what you lose is professional standing.

十Customer FAQ replies

For retail, service and B2B companies asked the same questions dozens of times a day.

Effort: Write up the thirty most common questions with standard answers; start with internal assistance (AI drafts, staff send), then consider automated replies. Two to four weeks.
Payback: Within a month. Response times drop and staff have time for the genuinely complex cases.
Risk: A wrong automated reply to a customer is expensive; go internal first and automate only once accuracy is stable.

Not sure which one to start with?

The two-minute self-check suggests a starting point for your situation — or just tell us where you are.

Take the self-check