Automating with AI without losing the human touch: concrete cases
A client writes at eleven p.m. and gets an accurate answer, in your tone. The next morning, the delicate question waits for a person, not a machine.
5 min readA high-end business can hand AI the first draft and keep the decision: answering frequent questions, sorting requests, summarizing a file, drafting a reply the team reviews. The human touch stays where it matters, and Quebec’s Law 25 requires informing people when a decision about them is based solely on automated processing.
This article is for information only and is not legal advice.
AI doesn’t need to replace anyone to be useful. Well placed, it takes away what keeps your team from clients. Here are four cases, and the rule that frames them. Our AI automation page describes our method.
Everyday questions
Hours, timelines, documents to provide, how a mandate unfolds. A chat on your site answers from your own information, at any hour. When a question goes beyond its scope, it hands over to a person, and says so.
An inbox, already sorted
An email, a form, a message: each request is classified, summarized in two lines and routed to the right person. Your team starts the day with what matters, not with an inbox.
Twenty pages, in five lines
A long email thread, a twenty-page file, a meeting: AI pulls out the key points, dates and decisions. The person responsible checks, then acts.
The draft, the signature
For a common request, AI writes a draft in your tone. Your team reviews it, adjusts it and sends it. The client gets a faster answer, written and signed by someone.
What stays in human hands
| Situation | Who decides |
|---|---|
| A simple, frequent question | AI can answer, with a human handover |
| A request to route | AI suggests, the team confirms if needed |
| A reply that commits the business | Always a person |
| An unhappy client or a delicate situation | Always a person |
| A decision about a client (accept, decline, prioritize) | A person, or the client must be informed |
The rule to know
When a decision about a person is based solely on automated processing, the business must inform them, tell them what information was used, and let them have the decision reviewed by a person. The simplest way to follow this rule: make sure a person keeps the decision.
You also need to know what is sent to the AI: only the information needed, to a provider that doesn’t reuse it to train its models.
A first case, well chosen
- List the questions your team answers most often.
- Gather the right answers: website, policies, documents.
- Test with real questions, reviewing everything at first.
- Widen the scope only once answers are reliable.
FAQ
Do clients know they’re talking to AI? They should. A chat introduces itself as such and offers to reach a person.
Can AI get things wrong? Yes. That’s why it answers from your information, and a person approves what matters.
Do we need lots of data to start? No. Your answers to frequent questions and your documents are often enough.
AI prepares; your team decides. Tell us which question you answer ten times a day.