What actually got automated, precisely
Be exact about this, because the vague version is either paralysing or dismissable. AI took the mechanical layer: transcription, captions and subtitle timing, rough assemblies, silence removal, reformatting one edit into vertical and square, standard colour and audio cleanup, background removal, generic stock b-roll and templated motion graphics. It did not take story judgement, pacing, knowing which take carries the emotion, translating a client saying “make it pop” into an actual change, or being answerable for the version that ships. The full role assessment is in will AI replace video editors.
The problem is the shape of that split. The automated half is exactly what juniors were paid to do, and it was also how editors learned. You developed an eye by logging hundreds of hours of footage and cutting rough assemblies badly until you cut them well. Remove the paid apprenticeship and the top of the profession stays healthy while the bottom rung disappears. If you are established, this is an efficiency story. If you are three years in, it is a wall.
You are more technical than you have been told
The single biggest blocker for editors considering this is a self-image problem, so take the audit seriously.
A timeline is a dependency graph. Change something upstream and everything downstream moves. You already think in the structure that trips people up when they meet functions, state and data flow for the first time. You invented version control before anyone explained it to you, which is why your folders contain final, final_v2 and final_v3_ACTUAL. Git is that instinct done properly, and editors tend to grasp why it exists faster than graduates do.
You debug under pressure. A render fails at 2am, the error is useless, and you work backwards through codecs, plugins, cache and corrupt media until you find it. That is the actual working experience of software engineering, and it is the part that makes most beginners quit. You do it routinely and call it Tuesday.
And the one worth the most: you have taste. Software teams are full of people who can build and short of people who can tell whether the result feels right, whether the pacing of an onboarding flow is wrong, whether an interaction lands. That judgement is genuinely hard to hire for and it is the thing you have spent your career developing. If you have ever wondered whether you are smart enough for tech, the honest answer is that the industry needs your half more than it needs another person who can write a loop.
Three lanes, and one of them keeps you in the industry
Lane one: creative tooling and pipelines. The bottleneck in creative production moved from making frames to building the systems that make them. Automation pipelines, asset management, render orchestration, internal tools for creative teams, and the fast-growing tooling layer around generative video. Every studio and agency is full of manual processes nobody automated because nobody there can build. This lane converts your entire career into a qualification rather than a thing you abandoned.
Lane two: product work. Design engineering, front-end with a strong visual sense, product roles where taste is the differentiator. Editors do unusually well here because they arrive with an opinion about how things should feel and the vocabulary to defend it.
Lane three: the full switch. General software development, where the creative background becomes an advantage rather than the job description. Most open roles, highest ceiling, entering at roughly RM 6,000 to RM 9,000 a month at the AI-capable junior level (the data), with full bands in the Malaysian salary guide.
The plan, and what to build
Six to twelve months, 10 to 15 protected hours a week, alongside whatever editing work you keep. The scheduling problem is the one that actually defeats people, not the material.
Build the tools your own trade needed and never got. A batch renamer that handles the naming convention your studio actually uses. A review tool that collects timestamped client notes without another subscription. A dashboard that tracks which cuts performed. A script that turns one master edit into every delivery spec a client asks for. These interview far better than a generic to-do app, because they prove two things at once: you can build, and you understand a real domain well enough to have opinions about it. That combination is what employers are short of.
All three lanes rest on the same foundation: being able to structure a problem, build, debug, ship and operate real software, with AI fluency and the judgement to catch what the model gets wrong. None of that learning is wasted whichever lane you pick, which is why it is worth starting before you have decided. The cheapest way to find out whether building suits you is the free trial: a week of real projects, a live instructor session, no card.
