Turn a written job ad into a short video that shows what the work looks like day to day. Pair a presenter avatar and a clear voice with footage of the real workplace, and say the pay and conditions out loud so candidates know what they are applying for.
Most job ads read the same: a list of duties, a list of perks, a line about being a fast-moving team. A short video can answer what the text never does, which is what an ordinary day in the role looks like. AI makes that video cheap enough to produce for every opening, as long as it shows the real job and says the pay and conditions plainly.
A video for one opening used to mean a camera, a presenter, an editor and a week of scheduling, so only the biggest employers bothered, and only for the roles they struggled to fill. Everyone else posted text and hoped. Generative video moves that cost down to an afternoon: you already have the ad, and the ad is the script.
The practical gain is not polish, it is coverage. A small company hiring a line cook, a warehouse picker and a bookkeeper in the same month can give each role its own clip, in the right language, without pulling anyone off the floor to be filmed. A recruiter who writes the same screening answers fifty times can turn the best ones into a short video once and link it from every posting.
Picasso IA puts the pieces in one place: presenter avatars, voice models, video generators and video editors, 502 models in all, which you can browse in the full catalog. New accounts get free credits, enough to find out whether the format suits your roles before you spend anything.
The title of this page is also the rule. A candidate who watches your video and then walks into a different job on day one has been misled, and with a video the mismatch is harder to explain away than with a vague paragraph. So decide first what the clip must show truthfully: the place, the tasks, the hours, the people they will work with, and what is hard about the job.
That decides where AI belongs. An avatar or a synthetic voice is a perfectly good way to deliver the spoken overview, the same way a recorded phone greeting delivers a message without a person in the room. Generated footage of a gleaming open-plan office is another matter if your team works in a depot, a clinic or a kitchen. Use your own photos and phone clips for anything that stands in for the workplace, and keep generated b-roll for context that is clearly illustrative, such as a visual for a software tool or a generic city shot behind a location mention. The AI b-roll use case covers that side in more detail.
Say it on screen too. If the presenter is an avatar, a short line at the start or a note in the description is enough. People forgive a digital presenter far more easily than a digital pretence of a human manager they will never meet.
Different roles need different clips. A table makes the choice quicker than prose, because the real question is what the viewer needs to see, not which tool is newest.
| Format | What it shows | Good for | Watch out for |
|---|---|---|---|
| Avatar presenter plus voice | A spoken overview of the role, pay and next steps | Remote or office roles that are mostly described, not seen | Reads as cold if it is the only thing on screen |
| Your own photo plus your recorded audio | The actual hiring manager speaking in their own words | Small teams where the manager is the draw | Use only your own face, with your consent, never someone else's |
| Real footage plus AI voiceover and captions | The real workplace and the real tasks | Trades, retail, hospitality, healthcare support, logistics | Needs a few honest phone clips to start with |
| AI b-roll as illustration | Context that cannot be filmed, such as a tool or a city | Software, field, or multi-site roles | Never pass it off as the actual site or crew |
| Short vertical cut | One question answered, such as what the shift looks like | Job boards and social feeds read on phones | Skipping the pay line because it is short |
Most good recruitment clips are the third row with the first row's voice on top: real pictures, a calm explanation, captions for people watching without sound.
This is the workflow that ends with a clip you can post, built on pages and models that exist in the toolkit. It assumes you have a finished job ad and a handful of real photos or phone clips.
The line candidates remember is the one that tells them what the job pays and what it asks of them. It is also the line most videos skip, because it feels awkward on camera. An avatar removes the awkwardness, which is one of the better arguments for using one: it will read a pay range with exactly the same tone as a greeting.
State what is true and complete. If the pay is a range, say the range and what moves a person inside it. If the schedule includes nights, weekends or on-call periods, say so. If the job involves heavy lifting, difficult customers or a long commute to a remote site, that belongs in the clip as well; it protects you from a wave of applicants who leave in week two.
A job video that hides the hard parts does not reduce churn, it moves it to the first month. Put the pay, the hours and the one thing people complain about in the first half of the clip, and let the rest be the pitch.
Pay disclosure rules differ by country, region and job board, and some places require ranges in postings. Treat this page as general information, not legal advice: check the rules where you hire, and have someone qualified, such as an employment lawyer or an HR professional, review the wording of anything you publish.
An honest page says where the tool stops working. For recruitment video there are five places:
None of these is a reason to skip the format. They are the reasons the first and last words of the video should come from a person who owns the posting.
Yes, in pieces. A job description makes a good script, an avatar or voice model reads it, and a video model or your own footage supplies the pictures. Picasso IA lets you run those steps in the toolkit one after another. What the AI cannot supply is the truth: you still decide what the day looks like, what the job pays and what the hours are, and check the finished clip against the written posting.
Both work, for different reasons. A real recruiter or hiring manager is the more credible choice when the people are part of the pitch. An avatar is a sensible choice when you want a consistent, calm delivery across many roles, or nobody on the team wants to be on camera. If you use an avatar, say so. Candidates accept it far more readily than finding out later.
Only if it does not pretend to be your workplace. Generated footage of an office, shop floor or clinic is invented, so a candidate who sees it will assume it is your site. Use real photos and phone clips for the workplace, and keep generated shots for abstract context such as a software tool, a city or a concept. Anything that could mislead someone about the job, the place or the team should be real.
For a presenter reading a script, heygen/avatar-iv takes text, an avatar and a voice. For animating your own photo with your own audio, kwaivgi/kling-avatar-v2 does the lip-sync. Voice models such as minimax/speech-2.8-hd and google/gemini-3.1-flash-tts handle narration, and google/veo-3.1 or bytedance/seedance-2.0 can make illustrative b-roll. The catalog has 121 text-to-video models, so test a few against the same script.
You should, and the sooner in the clip the better. Say the range, what changes it, and how often it is paid. Some jurisdictions and job boards require pay information in postings, and the rules vary, so check the requirements where you hire and have the wording reviewed by someone qualified. An avatar reads the figures in a neutral tone, which helps if the pay line feels uncomfortable to say yourself.
Only with their clear agreement, and ideally their input on what they say. Do not generate a likeness of a real colleague, a former employee or anyone else who has not agreed, and do not borrow a stranger's face. If nobody wants to appear, use a stock avatar or a fictional presenter and be upfront that it is one. Your own photo and recording, with your consent, is fine.
Long enough to answer what the written posting cannot, and short enough that people on a phone finish it. A good test is whether every line is something a candidate would ask a recruiter anyway: what the day is, where the work happens, the pay, the hours, what is hard. Anything beyond that tends to be company boilerplate, which belongs on the careers page rather than in the clip.
It can. You can write the script once, translate it, and generate a voiceover for each language, with captions in the same language. Have a fluent speaker read the translation before it goes out, especially the pay and hours lines, because a slightly wrong word in a contract-sounding sentence causes real confusion. The voiceover use case covers that workflow in more detail.
No, and they should not. A recruitment video describes the job to the candidate, it does not judge the candidate. Using generated video or voice to evaluate people raises fairness and legal questions that a qualified professional should look at, and in many places it is regulated. Keep the human decision on every application, and use the video to help the right people choose to apply.
Each generation is paid in credits, and the cost depends on the model and the settings you pick, so a number written here would be wrong soon. New accounts get free credits to try the workflow on one role, and current plans are on the pricing page, which is the only place worth trusting for figures.
Your next opening already has the script. Open the Picasso IA toolkit and turn the job ad into a clip that shows the real role.
Choose your language
