A careers page has a simple job: show who you are, list the open positions, and make applying easy. Yet many companies still have a PDF, a line on the main site, or a page nobody dares touch because “the agency made it”.
This guide shows how someone in HR, with no technical background, can build a real careers page with the AI they already use, put it online on the company’s domain, and keep it up to date afterwards without asking anyone. Count an afternoon for the first version.
Before you start: gather three things
You will go faster with these at hand:
- The open jobs, even rough: title, place, type of contract, a few lines on the role.
- Where applications should go: an email address, and the name of your HR or recruiting tool if you use one.
- A few photos of the team or the workplace, and your logo.
You do not need the texts to be perfect. You will be able to change every word later, without the AI.
Step 1: connect your AI to Leaf
Leaf works with the AI you already have: Claude, ChatGPT, Mistral, Cursor and others. Create a free account, and Leaf shows you an address to paste into your AI’s settings, with the steps for the one you picked. It takes a minute, once.
From then on, your AI can build in your Leaf space, with your rights. It acts for you, not on its own.
Step 2: answer the first-run guide
The first time, Leaf asks a few questions. All of them are optional:
- What do you want to make? Pick “A careers page”.
- Describe it in a few words, as you would to a colleague: “Our careers page: the team, our values, the open jobs, and an application form.”
- Who is it for, and what should they do? Candidates, who should apply.
- What should it do? Tick “Texts I edit myself” (so you can manage the jobs), “A contact form”, and “Send contacts to our tools” if applications should reach your HR tool.
- What look? Sober, warm, bold or playful, plus a site you like or your colours: “like our main site, dark blue logo”.
- What is it called? “Atelier Nord Careers”, for example.
Leaf turns your answers into a short brief and gives you one message to paste into your AI. That message tells it to open your brief and ask you what is unclear before writing anything.
Step 3: let the AI ask its questions
A good AI does not guess. Expect questions like:
- How many jobs are open now, and do you want one page per job?
- Should the form ask for a CV link, a cover letter, an availability date?
- Which language or languages?
- Do you have a domain in mind, like
careers.your-company.com?
Answer in plain words. If you do not know, say so: it will suggest something sensible. This is where the result gets good, so take five minutes.
What to ask for explicitly
A few things make a careers page better and are easy to forget:
- Jobs as editable entries. Ask: “Make the job offers something I can add and edit myself in Leaf, with a title, place, contract, description and a published or not status.” Leaf then gives you an editor made for job offers.
- One page per job, with its own address. Easier to share on LinkedIn and found by search engines.
- A short form. Name, email, a link to a CV or LinkedIn profile, a message. Long forms lose candidates.
- A clear confirmation. “Thanks, we answer within a week” is better than nothing.
- The phone. Many candidates look at job offers on a phone. Ask the AI to check it.
Step 4: look at the preview
Your AI builds the site on Leaf, then makes a preview: a copy at its own address, hidden from search engines. Before showing it to you, it checks its own work: screenshots on a phone and a computer, the form sent in test mode, errors read and fixed.
Open the preview yourself. Check:
| What to look at | Why |
|---|---|
| The jobs, their titles and places | They are what candidates search for |
| The form, filled in from your phone | It is where you lose people |
| The tone of the texts | It should sound like your company, not a template |
| The links back to your main site | Candidates want to know more about you |
Something is off? Just say it to your AI: “The intro is too long”, “Use our green”, “Put the jobs first”. It makes a new preview. You can also send the preview’s address to a colleague or your manager for a second look.
Step 5: approve, and it is online
When it is right, a person approves and puts it online. On Leaf this is the default for every site: the AI prepares, a human decides. The site is online everywhere within about a minute.
To use your company’s address, add the domain in the site’s settings. Leaf shows the one or two records to add where your domain is managed (often your IT team or agency handles this), then makes the security certificate by itself. Custom domains come with paid plans; until then the site has a free Leaf address.
If anything goes wrong after a change, the previous version comes back in a click.
Step 6: applications reach the right place
Each application sent through the form is:
- kept in Leaf, so nothing is lost;
- filtered against robots and floods of submissions;
- announced by email to the people who manage the site;
- and, if you asked for it, sent to your HR tool.
For that last part, your AI writes the connection, and asks for your HR tool’s key. You never paste that key in the chat: a person types it in Leaf’s settings, where it is sealed. Nobody can read it back afterwards, not even the AI. Your AI only knows the key exists.
Step 7: keep the jobs up to date, without the AI
This is where most careers pages die: six months later, closed positions are still listed. With Leaf, the jobs are entries you edit in a simple editor:
- Add a job: fill in the title, place, contract and description.
- Close a job: unpublish it.
- Fix a typo: change the text and save.
Changes are not online straight away. Leaf shows that the site has changes waiting, and you click “Update the site” when you are ready. You can preview the drafts first. If two people edit the same job, Leaf warns the second one instead of overwriting the first.
Working with an agency or a freelance recruiter? Invite them as an editor of this site only: they edit the jobs and texts, and see nothing else.
Going further
Once the basics are online, a few things are easy to add by asking your AI:
- Several languages, with the translations editable by your team.
- A short summary of each application written by AI for the recruiter, inside the tool.
- A weekly report of applications, sent every Monday.
- Visitors per job, to see which offers attract people.
Each of these is a sentence to your AI, a preview to look at, and an approval.
In short
- Gather the jobs, where applications go, and a few photos.
- Connect your AI to Leaf, answer the guide, paste the message.
- Answer the AI’s questions, and ask for jobs you can edit yourself.
- Review the preview on a phone and a computer.
- Approve. Add your domain.
- Applications reach your inbox and your HR tool.
- Edit jobs yourself, then click “Update the site”.
An afternoon for the first version, a few minutes for each change after that. And nobody has to say “send me the link” while you look for a localhost that only works on your computer.
Aussi en français : Une page carrières avec votre IA en une après-midi