Training
I run two course series in Singapore, both as small live cohorts of two days. The Agentic AI Series teaches you to build and run your own AI agents — a website builder, a content creator and a filmmaker. The AI Creative Series is the craft itself: social video, product content that sells, and cinematic short films. Team training runs on your own project, at your studio or online.
In practice: two series, two days each, in small live cohorts. Build and run your own AI agents in the Agentic AI Series, or work on the craft in the AI Creative Series — social video, product content that sells, cinematic short films. Team training runs on your own project, at your studio or online, and dates and fees are given per run, so ask.
Agentic AI Series — Website Builder, Content Creator and Filmmaker. Two days each, and you leave with your own working AI agent plus the workflow it runs on.
AI Creative Series — Video Lab for social media, Commerce Lab for product content that sells, Film Lab for cinematic storytelling. Two days each, and you leave with finished pieces, not just notes.
Dates are set per run rather than published as a calendar, so the quickest way to get yours is to ask on WhatsApp.
New to all of it, start with the AI Creative Series: Film Lab if you want to tell stories, Video Lab if you want social video, Commerce Lab if you sell a product. If your problem is doing the same work over and over, that is the Agentic AI Series — Website Builder, Content Creator or Filmmaker, whichever you repeat the most.
You can also see which guide sits with each course before you decide — the guides are the method, the courses are where you run it on your own material.
Ask whether you leave with a method you can repeat or a folder of clips you cannot explain. Ask whether you work on your own material, and whether you finish something.
Ask to see the teacher's own work, and look at the cohort size — a room small enough for someone to tell you honestly what is not working is worth more than a longer syllabus.
The open cohorts run in Singapore. Team training can be delivered at your studio or online, because it is built around your project rather than a fixed syllabus.
Both. Private team training is scoped to what your team already does, uses your own project as the material, leaves a documented prompt library behind, and includes a follow-up review session after two weeks.
The aim is a team that can repeat the pipeline without me in the room.
Send a brief through the work-with-me page or use the contact form, and say which course and which date you are looking at. If you are not sure which one fits, describe the work you want to be able to do and I will tell you honestly which of them does it.
The question worth asking of any course, ours included, is whether you leave with a method you can repeat or a folder of clips you cannot explain. Ask to see the teacher's own work, ask whose material you work on, and look at the cohort size — a room small enough for someone to tell you honestly what is not working beats a longer syllabus.
We would rather tell someone that ours is the wrong course for them than sell them the nearest one. The guides on this site are the method, free to read; the courses are where you run it on your own material.
Where this is taught
AI Film Lab: Cinematic Storytelling — 2 Days, taught live in a small cohort. Ask on WhatsApp (+65 9188 6948) for the next dates.
Keep reading
Process
Every piece I make runs through the same six stages: script, character bible, style lock, keyframe, motion, master. The order is the point. Each stage exists to make the next one cheaper, because the expensive thing in AI film is not generating — it is generating again.
Read →Method
Four rules carry every prompt I write: point at the attachment, name what cannot change, keep one thing per paragraph, and maintain a running negative sheet. They are not about clever wording — they are about removing the ways a model can misread you.
Read →Method
Consistency is the hardest problem in AI film: the same character, the same world and the same look in every single shot. It is solved with references and diagnosis, not with better luck — five things do most of the work.
Read →Diagnosis
AI video gives itself away in six ways: the face drifts, the physics float, the light is flat, the camera moves like no camera, the sound does not belong, and small details mutate between frames. All six are failures of preparation rather than of the model, which is why each one is prevented before generation instead of repaired after it.
Read →Definition
A style lock is one approved reference sheet that governs how every frame of a film is rendered: the light, the palette, the amount of detail, and the way the camera behaves. You approve it once, before animation starts, and then point at it on every shot instead of re-deciding the look shot by shot.
Read →Definition
A shot list is the film written down as shots before anything is generated — what each shot is for, the emotion it has to land, how long it lasts, and how the camera behaves. It matters more in AI than on a set, because changing your mind in words costs nothing and changing it after a render costs a re-render.
Read →Getting started
No — neither is required to start, but the work still asks you to make the decisions those skills exist for: what the shot is for, where the camera stands, and what must not change. The tools remove the cost of a shoot; they do not remove the need for a plan, and the planning is teachable.
Read →Commercial work
A product film is a short list of shots with one job each: a hero shot that shows the thing itself, a detail shot for texture and proof, a context shot that shows who it is for, and one line that carries the offer. It is short, because a launch film is watched on a phone between other things.
Read →