TL;DR
- The first question for a training series isn't who makes it. It's how often the content will change. Content that changes every quarter shouldn't be locked into footage that costs a shoot day to update.
- Film the parts people have to believe: why the company exists, what leadership expects, a supervisor explaining why a safety rule matters, and demonstrations on your real equipment.
- An AI presenter suits training that delivers information and gets revised: SOPs, policy refreshers, yearly compliance modules, and anything needed in more than one language.
- Training on how to use software is usually a screen recording. Filming a person talking about a system is slower and less clear than showing the screen.
- A programme can be mixed: one shoot for the parts that rarely change, with everything that gets revised kept somewhere it can be edited.
Someone in HR has been told to get "a set of onboarding videos" done this quarter. The new system goes live in two months, the SOPs were rewritten in the spring, and the yearly compliance refresher is due again. There's no in-house video team.
The usual first move is to find someone who can make the videos. The more useful first question is this: how long will each of these videos stay correct?
A welcome from the managing director might hold for three years. A module on the claims process might be wrong by the next policy update. A walkthrough of the new HR system will be wrong as soon as the vendor moves a button. They're all "training videos", but they have very different lifespans, and the lifespan should decide how each one is made.
What are the three ways to make a training video?
There are three, and each one is good at a different part of a training programme.
Film it with real people. This is a proper shoot: crew, lights, your premises, your people. It's what you use when the viewer has to believe what they're watching. A founder explaining why the company exists. A head of operations saying what's expected. A supervisor on the floor explaining why a safety rule exists, in a voice the staff already know. A demonstration of a task on your actual machine, in your actual warehouse, because the trainee has to recognise it on day one.
Use an AI presenter. A generated presenter reads a script you approve. It's good when the job is delivering information rather than asking to be believed: SOP walkthroughs, policy refreshers, the yearly compliance module, product knowledge for new sales staff. The advantage is maintenance. When a step changes, you change the script. You don't rebook a shoot. When you need the same module in several languages, the extra languages don't each become a separate production.
The trade-off is trust. An AI presenter costs some credibility with the viewer, which is fine for "here is how to submit a claim" and wrong for "here is why we care about safety". We've written about that judgement separately, in AI presenters and digital spokespeople and in who should be on camera. This page doesn't repeat it.
Screen-record it. If the training is "how to use the system", the screen is the training. A recording of someone actually doing the task, with a clear voiceover and the clicks visible, teaches better than a presenter describing the steps. It's also the cheapest of the three, and often something your own team can make with a screen recorder and a decent microphone.
If you'd rather hand it off, we take these on too: the screen capture, the voiceover, the edit and the captions.
Which training videos aren't worth RM15,000 to film?
A filmed corporate piece with us starts at RM15,000. Training video is listed on our corporate video production page under that same band, made to the same standard as the outward-facing work. That's the right spend for some training. For a lot of it, it's the wrong spend. Here's where it usually doesn't pay:
- Anything that walks through software. The interface will change, and when it does the filmed version is out of date with no cheap fix. Screen-record it.
- SOPs that get revised more than once a year. If a step changes and the only fix is a reshoot, the video will soon be "mostly right", and staff learn to ignore it.
- Policy and compliance readouts. If the content is a set of rules being read out clearly, a filmed presenter adds cost without adding much trust. An AI presenter or a narrated slide module does the same job.
- The same module in four languages. Filming each language version separately multiplies the shoot. This is the kind of job the AI route is for; on our AI video production page it starts at RM4,500.
None of this means training should be cheap across the board. It means the shoot budget should go to the parts that can't be made any other way. That's the same point as the note on lighter productions in what a corporate video costs, applied to a whole training programme instead of one video.
When is an AI presenter good enough for staff training?
When the module is information that has to be accurate, consistent, and easy to update, and the viewer doesn't need to trust the presenter as a person.
Two things decide whether that holds for a particular programme:
How an update actually works. "Changing the script" sounds instant. It still involves steps: someone on your side approves the new wording, the changed section is regenerated, and the module is re-exported to wherever your staff watch it. The facts still need sign-off. A generated presenter doesn't move responsibility for a wrong instruction off the company that issued it.
With us, each change like this is quoted on its own as a small job: you send the approved new wording, we regenerate the section it affects and re-export the module. Because it's charged per change, it's worth batching edits that arrive close together rather than sending them one at a time.
Which languages, and who checks them. Many workplaces in Malaysia run in more than one language. A module that exists in English only may not reach the people who most need it.
With us, AI-presenter modules have gone out in English, Bahasa Malaysia and Chinese. Each language version is checked by someone fluent on your side before it's released. You know your own terms and your own staff, so that sign-off stays with you.
Can you film part of it and use an AI presenter for the rest?
Yes. For an onboarding series it's often the most sensible shape, and it's the structure we'd put to an HR team planning one.
The idea is to split the programme by lifespan, not by module:
- One shoot day for the parts that rarely change. The leadership welcome. The "why we do this" pieces. Demonstrations on real equipment and in real locations. A tour of the premises for new joiners. These are what a camera is for, and they won't need refilming when a form changes.
- An AI presenter for the parts that will be revised. Step-by-step SOPs, policy modules, the compliance refresher, and the language versions. These stay editable.
- Screen recordings for the systems. Made by whoever knows the system best, and re-recorded when the vendor updates the software.
The filmed parts give the programme the human voice that makes staff take it seriously. The AI parts keep it maintainable, so it doesn't go out of date the first time the SOPs are revised.
How to read a quote for it. Ask for the two halves to be priced separately: the shoot day as filmed work, and the AI modules on their own. A single package number hides which part you're paying for, and separate lines let you add or drop modules later without renegotiating the shoot.
Whose face the AI presenter has. Usually it's one of your own people. We film your trainer or HR lead once and build the AI presenter from that footage, so staff are learning from a face they already know, which feels a lot less like a generic module. The question to settle up front is what happens if that person leaves. Our advice is to put it in writing before the shoot: a consent from the trainer that grants the company use of their AI presenter, and a clear promise from the company that it will only ever be used for internal training material, nowhere else. Trainers are far more willing to sign when that limit is spelled out.
It also shortens the shoot. Building the presenter takes about half a day of the trainer's time. Filming the same trainer for a full series used to take us a full day, sometimes two.
So which one does your training need?
Run each module through three questions: how long it will stay correct, whether the viewer has to believe the person speaking, and how many languages it's needed in. Whether it shows a screen or a physical place is usually obvious.
| The module | Stays correct for | Has to be believed? | Languages | Usually best made as |
|---|---|---|---|---|
| Leadership welcome, values, "why we exist" | Years | Yes | One, maybe two | Filmed |
| Safety: why the rule matters | Years | Yes | One or more | Filmed, with a real supervisor |
| Task demonstration on your own equipment or site | Until the equipment or layout changes | It has to be recognisable | One or more | Filmed, with other languages added as voiceover |
| SOP walkthrough, step by step | Until the next revision | No, it has to be accurate | Often several | AI presenter |
| Policy and compliance refresher | A year, often less | No | Often several | AI presenter |
| How to use a software system | Until the next interface update | No | One or more | Screen recording |
Two things change the answer:
- If a module is shown to customers or partners as well as staff, it's no longer just training. It's part of how the company presents itself, and that tips it towards filming. The reasoning is in when not to use AI video.
- If nobody will own the updates, the AI route loses its main advantage. An AI module that's never updated just goes out of date at a lower cost. Decide who on your side signs off a change before you choose the format.
What this page can't settle
It can't tell you what your training should say. The content, the SOPs and the compliance requirements are yours, and so is the sign-off on whether each instruction is correct. We can make it clear, watchable and easy to maintain. We can't make it accurate if it wasn't to begin with.
It also doesn't cover your learning platform. Whether your LMS needs plain video files, subtitle files or a packaged course format is something to confirm with whoever runs it before production starts.
For reference, what we hand over for training work is MP4 files plus a separate subtitle file for each language. We haven't delivered packaged course formats such as SCORM, so if your platform needs one, settle that before production starts.
If you're planning a programme like this, the corporate video production page is where to start the conversation. Bring the list of modules and a rough idea of how often each one changes. That list will decide most of the plan.
Thinking about a project like this?
Tell us what it needs to do — we'll take it from there.
Frequently asked questions
- Should a company film its training videos or use an AI presenter?
- It depends on what each module is doing. Parts that need to be believed, like a leader explaining the company's values or a supervisor explaining why a safety rule exists, are worth filming with real people. Parts that deliver information and get revised, like SOPs and policy refreshers, suit an AI presenter, because changing a line becomes a script edit rather than a reshoot.
- Is a screen recording good enough for software training?
- Usually, yes. If staff are learning a system, the screen is the content. A clear recording with a steady voiceover shows the clicks better than a filmed presenter describing them. It also goes out of date whenever the software's interface changes, which is one more reason not to spend a shoot day on it.
- How do you decide which training videos are worth a professional shoot?
- Ask three things about each module: how long it will stay correct, whether the viewer has to believe the person speaking, and how many languages it is needed in. Long-lasting content that needs to be believed is worth filming. Content that changes often or exists in several languages usually isn't.
Keep reading
- What Is a Brand Film (and When a Corporate Video Isn't Enough)?A brand film goes after belief, not information. Here's what actually makes one, how it differs from a corporate video, and the moment a straight capability video stops being enough.
- Corporate Video, Brand Film, TVC, Explainer: Which One Do You Actually Need?Four kinds of company video get lumped together and briefed as if they're interchangeable. They're not. Here's what each one is for, and how to pick the right one before you spend.
- Product Videos That Sell: a Guide for Malaysian Consumer BrandsA product video isn't a trust asset — it has one job: make someone want the thing and know why. Here's how to brief one that actually sells, for Malaysian consumer brands.