It started with a question I prompted on Claude: If I am forced to direct my film myself, how do I prepare? Because I will be surrounded by people who know the craft better than me.

I asked that question because I am not proud of what I feel is my shortcoming.

I don’t want to make what I don’t know become the subject of scrutiny by people who know a lot more than I do. That’s the specific fear. Not imposter syndrome in the way that phrase gets used until it means nothing. Something more precise than that. The fear of being caught not knowing in a room full of people whose entire identity is built on knowing. The fear that the gap between what I’ve built on the page and what I can execute on the floor will be visible to everyone before I open my mouth.

It doesn’t help that I don’t identify myself as a film buff. Ask me about a film that isn’t a Hollywood blockbuster and a cat gets my tongue. I don’t have the references. I don’t have the lineage. I can’t say this scene is Tarkovsky or the framing here is pure Wong Kar-wai — the shorthand that filmmakers use to signal to each other that they belong in the room. In a room full of people who can, that absence is visible in a way that’s hard to hide and harder to explain.

What I have is literature. Theatre. Twenty years of reading character from the inside out, understanding what a text withholds, knowing that the space between words is where the real information lives. A completely different formation. Not inferior. Just built from a different direction.

I didn’t know yet whether that was enough. So I asked.

What came back wasn’t a preparation plan. It was a process — one that took several sessions, and eventually led back into the script itself. Into lines I had already written without fully understanding what they were.

The Clues on Hindsight

A theatre director friend was the first external confirmation. He got unreasonably excited about an action line.

Not a plot point. Not a piece of dialogue. An action line.

“PARANG keeps his gaze on the KHUKRI, too hot to hold, too rude to throw away.”

He said that this is the line that will make actors dive deep. He wasn’t talking about the story. He was talking about what the line does to a performer — the way it hands them a problem rather than an instruction, a friction rather than a feeling, and trusts them to find what lives inside it.

I wrote it without thinking about any of that. I was writing a character in a specific moment. I wasn’t thinking about actors at all.

But he was right. And his being right meant something I wasn’t ready to fully absorb yet.

“His scream is silent. The music carries the pain of a man who finally knows what it means to break.”

My friend’s reaction to this one was different. The first line excited him because of what it asks of an actor. This one excited him because of what it refuses to ask of one — and what that refusal demands instead.

The most painful moment in the scene is explicitly handed elsewhere. The music carries the release. But that’s a post-production decision. On the day, on the floor, there is no music. There is only an actor and a camera and a director asking for something almost impossible to articulate: dig deep into the feeling of loss where you are a few steps too late. But know that no one is going to hear how you feel. They need to see it.

The silence isn’t emptiness. It’s the full weight of the break held inside the body, behind the eyes, in the jaw, in the hands — real enough that when the music arrives in the edit, it has something to carry. A performed silence is visible immediately. The audience sees someone acting quiet rather than someone breaking quietly.

So the line isn’t asking less of the performer. It’s asking something more specific and more difficult. Don’t release. Hold it until the music — which you will never hear on set — takes it from you.

Three collaborators — actor, composer, editor — each assigned their share of a single moment, without any of them doing each other’s work. I didn’t know I was doing that. I was trying to describe what the scene felt like from the inside.

A screenwriter describes. What I wrote was architecture. Those are not the same thing.

“She no longer wants PARANG to defeat him. She wants ADITYA for herself.”

This appears after Mayang’s brother is revealed for what he is. After everything she didn’t see, and then suddenly saw.

The line tells the actor nothing about what to show. It tells her everything about what to want. And those are completely different performances — because a want shapes the body before the body knows it.

“She rages” is one film. “She wants ADITYA for herself” is another. The second rage has a direction the first doesn’t. It isn’t grief moving outward in every direction at once. It’s something that has found its object — ownership, betrayal, twenty years of proximity to someone she thought she knew, and the specific fury of finally understanding what she was looking at. The audience won’t be able to name what’s different. They’ll feel that the fury is aimed. That it knows exactly where it’s going.

I wrote that line because I needed to know what Mayang was feeling in order to write what she did next. But in the process of answering my own question, I handed the answer to someone who would need it in a very different room, years from now, preparing for a scene they hadn’t shot yet.

The fourth line is the one that settled it for me.

“Her fingers claw the floor. Every ounce of energy that is left in her is about gaining every salvageable victory she can earn.”

Not survival. Not escape. “Salvageable victory”.

In a moment of complete violation, with nothing left, Mayang doesn’t collapse. She reorganises. The clawing fingers aren’t despair — they’re strategy. The smallest possible defiance, chosen with full intention.

Without that last line, an actor plays suffering. With it, she plays Mayang.

I wrote that line because I needed to understand the character. But what I actually wrote was a brief to a performer about who this person is at her most reduced — and the answer is: she is still deciding what the moment means for her. The violation doesn’t get to define her. She does.

That’s not an action line. That’s the most important thing I could tell the actor playing Mayang before she walks into that scene.

So Should I Direct?

Here is what I’ve been working out from these four lines.

Directing isn’t doing the actor’s job for them. It’s creating the conditions for them to do it themselves. There’s a fundamental difference between telling an actor what to feel and telling them what to carry. “Feel sad here” produces a performance of sadness. “What do you choose to feel about it”, and “What do you tell your daughter who doesn’t know” produces a specific human being making a specific choice in a specific moment.

I didn’t learn that from a directing seminar. I learned it from lines I wrote without knowing what they were — lines my friend recognised before I did, lines that kept doing something other than what scene directions are supposed to do.

The craft gap is real. I won’t pretend otherwise. There are things people who have spent their lives inside cinema know that I don’t, and won’t, and will need other people in the room to hold for me. That’s not going away.

But the four lines suggest the gap might not be where the film lives.

I walked in through the literature door. Through theatre. Through twenty years of learning how character operates from the inside, how a text withholds, how the space between words does more work than the words themselves. That’s a different formation from the one the room expects. It produced something I couldn’t have produced by studying film — because I wasn’t studying film. I was studying people.

The fear is still there. The bite just feels less painful.

What changed is the evidence. The directing was in the writing before I was willing to call myself a director. I just needed a question I wasn’t proud of, and enough sessions to find out what the answer already was.

I spent part of this week generating an image of the Khukri. Not a prop. Not a placeholder. The Khukri — the one that belongs to Parang, the character the trilogy is named after.

Four generations to get it right. What follows is what the process revealed — not all at once, but in stages, the way discoveries usually happen.

The brief I gave

For context, most of my prompts are AI generated. For this one specifically, the initial prompt was technically complete. Blade geometry, handle material, surface condition, lighting direction, camera framing. Everything a generation model needs to produce an accurate image of a kukri, so what got produced was an accurate image of a kukri.

Correct. Precise. Dead.

By the second generation, I asked whether the weapon should be given more personality. The response asked me to resolve “living” into generation language — what does living look like, visually? So I gave it a lore. The blood rule, based a myth tradition tied to the actual blade, because it felt right for Parang: draw the Khukri and it needs to be fed before you can sheath it. Fail to do that and it comes after you instead (Taming Sari, and other bladed weapons in history, has a similar trope).

That is not a visual instruction. That is the weapon’s nature. The specific truth of what it is carrying.

What came back was a prompt built on visual symptoms: wood grain running the wrong direction, a shadow that doesn’t match the light source, subtle wrongness only visible if you look. Technically inventive. But the lore had been translated into generation language — and in that translation, the story had been removed.

The AI was doing what it was built to do: Resolving the story into something the model could render, pushing the photorealistic output by converting interior truth into visible symptoms.

I took the lore back and wrote the prompt myself.

The fourth attempt — a different register

A fresh thread on the image generative AI. The geometry was still in there — deep forward belly, thick spine, cho notch at ricasso, brass bolster, dark hardwood. But this time the living quality was not translated. It went in directly, in the language it arrived in:

The oils from the skin contact of the last bearer is visible. But that person has not touched it anymore for reasons.

Blade: scratches from use. However, it is also clean. Too clean for the blade of this age. As if the Khukri is cleaning itself.

If you look carefully, it’s as if it’s asking for another life, and will come after you if you sheath it away without giving it what it wants.

The image that got generated was accepted immediately.

What the image says before I showed anyone the prompt

I shared the image back to my prompt generating AI for confirmation. The assessment came back before the prompt was shared:

The fuller shadow is the detail that makes it work. It should not be that dark given the lighting direction. It is not explained by the studio setup. Something is in there.

Then I showed the prompt. What my language had done that the technical drafts had not:

“My version said ‘too clean for a weapon that has seen use’ — yours said why it is too clean. Generation understood the difference.”

Describing a condition produces the condition. Naming the cause produces something that means it.

The difference between the two prompts was not craft. It was register. The technical draft described what the living quality might look like. My prompt told the generation what the weapon was — what it wanted, what it had lost, what it would do about it.

Generation cannot render intent literally. But it reads register. And it responded with a fuller khol shadow.

What the image kept revealing

Left: Parts of a standard kukri. Right: “PARANG! A Hikayat”canon-locked Khukri

A few exchanges after acceptance, I asked whether the extended khol was unusual.

A standard khol (groove) on a kukri runs perhaps a third of the way from the ghari (ricasso). This one ran almost the full blade length. That is not typical kukri geometry. The AI offered two possibilities — generation drift, or something that belonged to this weapon specifically.

I loved it. It implies the blade is heavy at the back. Strong spine through and through. Which means it hits hard. A weapon like this either bleeds something or breaks something. Either way, a weapon carrying a blood rule should not have the build of something that hesitates. It asks for returns. Someone or something will get incapacitated.

The extended fuller is now canonical geometry — locked.

Then the chirra (fuller). It moves like a serpent finding a path. To the point that the serpent “eats” into the cho (notch).

I did not prompt either of these details. They arrived because the prompt had told the generation what the weapon was carrying. The geometry answered.

Why the instinct had been correct all along

By this point I had run full or partial generation sprints on all eight main characters in PARANG!’s first film. The pattern that had been building — not as a rule, just as something that kept being true — was that the prompts with the highest acceptance rates were never the most technically complete ones.

They were the ones that told the model what needed to be felt.

Physical description places a character correctly. Race, age, build, the way a face has been lived in. That work matters. But the images that landed — accepted without hesitation — were the ones where the prompt also carried what was buried deep in the person’s soul. Their wound. What they were pretending not to carry. The thing they had decided a long time ago and never revisited.

When that was in the prompt, the face answered it.

Eight human subjects. The question I asked about the Khukri — should we give it more personality? — was the same recognition applied to an object. The weapon was a character. It had an interior.

The instinct was correct from the start. What the accepted image made visible was why.

F16

Every subject in this pipeline — character or object — has a surface and an interior.

Technical description constructs the surface. The interior is what makes the surface legible.

The interior is not mood language. It is not atmosphere added after the geometry is established. It is the specific truth of what this person or object is carrying — their wound, their hunger, their contradiction. That truth goes into the prompt directly, in plain language, at the register you would use to brief an actor.

This is not a prompting technique. It is a creative standard — and it is not new. It is the same standard a director applies when briefing a costume designer not just on period and rank but on what the character has survived. The same standard a production designer needs to build a set that carries the emotional register of a scene. The same reason a technically correct performance can leave an audience unmoved.

Technical information produces technical results. The interior was always the primary information.

The AI, left to its own instincts, will translate story into visual symptoms. It is doing what it is built to do. The conductor’s job is to hold the standard the orchestra will not reach on its own — and to know when to stop translating and go direct.

Where this goes

PARANG! is being built with AI in the pipeline. Not because it is cheaper. Because it is the right instrument for what we are trying to make.

Pipeline Protocol “F16” is now back-propagated across every prompt architecture in the pipeline — character references, weapon references, environment references, shot production prompts. Canon nature leads. Geometry follows.

The mistakes were not detours. Output “W2” and “W3” failing is what made “F16” visible. The instinct, the wrong turns, the moment the image landed and the understanding arrived after — all of it is part of the learning protocol. You audit not because the first version was bad but because the standard only becomes clear through the work of getting there.

The Khukri took four generations. That is exactly how many it needed.

PARANG! A Hikayat is the first film in the PARANG! trilogy. Now in development.

Antara Deria is the Substack of Armen Rizal Rahman — writer, creator, and founder of Antra Indra Productions.

Twenty years ago, I received a story.

Not the full thing — a seed. A world glimpsed at an angle. The kind of concept that settles into you quietly and refuses to leave, that you carry around for years without quite knowing what to do with it, that grows in the dark without your permission.

What started as a random joke based on a popular film became something else.

PARANG! — a 13th century Nusantara action-horror set in a world where silat masters walk alongside the supernatural, where iron nails became instruments of control, desire and corruption, where the horror is not the creature but what the creature reveals about the human beings around it.

I knew what it was. I couldn’t build it. Not alone, and not in the way the traditional model demanded: the team, the funding, the development executive, the two-year timeline, the compromise.

So for twenty years, I kept it. And then, six months ago, everything changed.

What One Subscription Produced

Then in 2025, within six months of working alone, with no production background and no formal film training, I produced:

A completed feature screenplay. A full trilogy architecture across three films. A world canon at version 16 (at the time of this writing) — governing everything from cosmological rules to period-accurate dialogue register to what the world is not. A visual identity. Character portraits. Key art. Online visual identity, public facing and live. A pitch deck ready for co-production enquiries. A proxy audience test rating the screenplay at 8.5 out of 10 before a single frame was shot.

When I tell people this, they assume I mean I used AI to generate all of it. That’s not what happened. That’s not even close to what happened.

What I actually did was learn to conduct.

The Distinction That Changes Everything

A conductor doesn’t play the instruments. They don’t compose the piece. What they do is hold the standard — the interpretation, the vision, the non-negotiable sense of what this should become — and shape every execution toward it.

The orchestra can play the notes. Most orchestras can play the notes. The conductor is what determines whether you get sound or music.

This is how PARANG! was built. Not AI-generated. AI-conducted.

The initial treatment was skeletal. I didn’t know how much I should be writing in a treatment. So I let AI do the writing on my behalf. About 80% of it was produced by AI. I even let it dictate the core emotion of the story. I had a basic plot sequence. To me that was enough.

It met with poor feedback. When I read back Treatment v1 today, I knew the feedback was warranted. But the feedback also excavated one of the best advice I received at that time: “Put in the rawest, immoral, unadulterated, uncensored, worst version of you into the story. Put in anything. But it has to be you.”

So I did the complete reverse of what I did with version 1. I put the flesh in raw — unfiltered, just thoughts to screen. The raw version of the story, where everything is completely me, is v2. I ran it through AI for analysis. That produced v2.1.

The seeds of a 20-year old story started to take form. But on top of it, the seeds of Conducting AI began to tap the conductor’s stand for attention.

From there: painstaking scene-by-scene rewriting, checked against canon. Problem areas surfaced, developed, pressure-tested. Major plot additions earned new version numbers. Research fed back into world logic. Every iteration vetted for internal consistency.

Draft. Judge. Rebuild. Pressure-test. Expand. Repeat.

As the cycle continues, I put more of my heart into the story. My own experiences hidden in the characters. The story became personal.

No one tells me how to be me, so I applied it as a standard to what comes out of each version.

Version by version. Scene by scene. Prompt by prompt. Always with the vision intact. Always with the standard non-negotiable.

The canon document at version 16 is not what version one looked like. It is the residue of thousands of decisions about what fits this world and what doesn’t — decisions that compound, that layer, that produce something no prompt could have generated and no AI could have authored.

That’s what twenty years of accumulated judgment looks like when you finally give it the right tools.

What the Process Revealed

The most important thing I learned wasn’t about AI. It was about the audit.

The loop is simple: Prompt → Output → Audit → Feed → Output → Audit. But the step everyone undervalues, the step most people using AI for creative work skip entirely, is the audit. Not reviewing the output. Auditing it — applying your standard, not just your preference. Knowing why something fails, not just that it does. Being able to articulate the rejection in the creator’s voice.

Most people using AI for creative work produce version one and stop because they cannot tell version one from ‘good’. Conducting AI runs to version 16 because the standard is non-negotiable.

Here’s what I also learned: the AI I used most consistently — Claude — is designed not to treat the audience as stupid. It pushed back. It rejected my weaker proposals on structural grounds, clearly and with reasoning. Three consecutive structural proposals about a character named Ayu, tested in a single session, all rejected through the AI’s own logic — and all three rejections confirmed as correct by me after the fact.

That’s not a tool. That’s a collaborator you’re conducting.

The Business Produced In Parallel

The six months didn’t just produce a film franchise.

It produced a named, documented, teachable process that compresses development timelines from years to months without compressing the depth of the work. An AI-chain methodology that is replicable across any concept, any language, any market.

It produced a business. Antra Indra Productions is an IP incubator — identifying undervalued Southeast Asian creative concepts at origin, extracting and building their latent world architecture systematically, and positioning them for maximum return before exiting or licensing. The methodology is the engine. PARANG! is the proof of concept.

And it produced a strategic architecture. How the team gets built. How the funding gets structured. How the IP gets protected. How the infrastructure — the methodology, the proprietary model being trained on twenty years of accumulated judgment — stays separate from the film investment, legally and operationally.

One person. One AI subscription. Six months. Not just a franchise. A company.

The Thing Nobody Talks About

AI, at its current state, has access to the collective wisdom of billions of human thoughts and decisions. But that wisdom appears in fine threads. Generic at the surface, extraordinary underneath — if you know how to look.

To spot those threads requires something that can’t be automated: judgment developed over time, across a specific domain, by a human who has been thinking about one thing long enough to know what the right answer looks like before it arrives.

Twenty years of carrying a world in your head is not wasted time. It’s not nostalgia. It’s the accumulation of a standard — a geological deposit of decisions about what this world accepts and what it rejects. That standard is what you conduct with.

The field is flat now. Tool access is equal. But the twenty years aren’t.

What This Means

PARANG! is not a pitch for a film. It is a demonstration.

A demonstration that the old development model — teams, funding, two years, high burn rate, compromise at every stage — is no longer the only path to a franchise-ready property. A demonstration that a single person with the right accumulated judgment and the right methodology can build something with structural integrity, commercial architecture, and genuine creative depth.

The methodology that produced it is called Conducting AI. It is named, documented, and teachable. It is replicable across any concept, any language, any market. It compresses development timelines from years to months without compressing the depth of the work.

But it requires something most people using AI don’t have and can’t acquire quickly: the capacity to recognise what the tool is actually offering — not the surface output, but the thread of genuine insight running underneath it — and build with it systematically until the gem drops.

That is what Conducting AI is. Everything else is infrastructure around that sentence.

PARANG! is in active development.

Antra Indra Productions — between senses

The numbers are no longer in dispute. Southeast Asian horror works. It crosses borders, builds audiences, and travels in ways that confounded early sceptics. Impetigore found international festival screens. The Medium unsettled viewers who had never set foot in Thailand. Sewu Dino broke streaming records domestically before anyone in a Western development office had learned to pronounce it correctly. The genre has proven itself.

Mainstream entertainment responded the way it always does when a market signal becomes undeniable. It leaned in. Platforms commissioned. Scouts arrived. The conversation shifted from “does SEA horror travel?” to “how much SEA horror can we move?”

But something is being missed. And the missing is structural, not incidental.

Horror-action is not a new idea. Blade made it cool. Resident Evil made it a franchise. Buffy the Vampire Slayer made it a cultural institution. Each of them built a fandom that endures precisely because the world architecture was total — the mythology had rules, the rules had consequences, and the story never cheated on either. The genre doesn’t just ask for good fight scenes and good scares. It asks for a world complete enough to sustain both simultaneously without either cancelling the other out.

That is a high bar. Most stories cannot clear it.

SEA action has had its moment too. Ong-Bak announced itself before The Raid existed — Muay Thai rendered with a ferocity that needed no translation. The Raid then confirmed it wasn’t an anomaly. Silat-driven cinema crossed over because the language of the body in extreme motion is universal in a way that dialogue never quite is.

Two proven genres. A template for their merger that has already demonstrated global franchise potential. And yet SEA has not produced a horror-action story strong enough to carry both.

The question worth asking is why.

The answer is not talent. The region has produced filmmakers of genuine vision across both genres independently. The answer is not audience appetite — the fandom for exactly this kind of story is demonstrably real and hungry for something new to attach to.

The answer is architecture. Specifically, the absence of development infrastructure willing to go deep enough into the source material to extract what’s actually there.

Because what is actually there is extraordinary. And it has been hiding in plain sight.

Before a Muay Thai fighter throws a single strike, he performs the Wai Kru Ram Muay — a ritual that traces its roots to ancient warriors seeking blessings from monks before battle. The ceremony is not warm-up. It is invitation. The fighter prepares his body and soul to reach what practitioners describe as a divine and supernatural condition — a state where the physical can draw on something beyond itself. The fight cannot begin until that connection is made.

A Muay Thai fighter’s Mongkol — the sacred headband worn into the ring — is blessed by a monk before it ever touches a fighter’s head. The Sak Yant tattoos covering the bodies of fighters across Thailand are applied by Buddhist monks with bamboo needles, each design a specific consecration. The warrior’s skin itself becomes a ritual object before combat begins.

In the Malay and Indonesian world, the bomoh and the dukun have governed the relationship between the visible and invisible for centuries. Their role spans healing, protection, and harm — drawing from the same well of animist, Hindu-Buddhist, and Islamic spiritual inheritance that shapes daily life across the archipelago. In parts of East Java, it is said that everybody is thought to be able to do a little magic. The spiritual is not a separate domain. It is the operating system.

And then there is the keris. The blade is not complete — not fully itself — until it has been through the semangat infusion ritual: washed with lime, smoked over burning incense, its spirit acknowledged and activated. The weapon and the ceremony are not separate things. The instrument of violence is also a sacred object. There is a belief, held across the region with varying specificity, that certain blades once drawn cannot be sheathed without tasting blood. Whether literally true is beside the point. The belief exists because it makes sense within a worldview where a weapon is not a tool but a participant. It has agency. Drawing it is an act with consequences that extend beyond the mechanical.

None of this is unusual within the region. None of it is fringe. These are not superstitions at the margins of otherwise rational cultures. They are the architecture. The tradition of drawing on forces beyond the self — to protect, to heal, to fight, to win — runs through every culture across this geography, expressed differently but rooted in the same understanding: the visible world is governed by something operating just beneath it, and those who learn to engage with that something operate at a different level than those who do not.

But here is what all of these traditions share, and what mainstream entertainment has not yet noticed.

Every one of them — silat kebatinan, the Wai Kru, the bomoh’s craft, the keris ritual — is oriented toward human conflict. Man against man. Community against community. The spiritual infrastructure exists to tip the scales in a human contest. Protection from a rival. Strength against an opponent. A favourable outcome in a dispute that was always, at its foundation, human.

The tradition assumed the threat was human.

What happens when it isn’t?

What happens when the warrior trained in a system built for human conflict is asked to fight something that exists outside that framework entirely — something that cannot be negotiated with, cannot be reasoned with, and was never human to begin with? What happens when the same spiritual infrastructure that was built to protect the community is the infrastructure being used against it — corrupted from within, turned into a weapon by someone who understood its mechanics well enough to invert them?

That is not a genre question. That is a civilisational one.

And it is a question the tradition has never been asked to answer — not at the scale that cinema can demand of it.

Blade works because the vampire mythology is total. Resident Evil works because the infection logic is internally consistent. Buffy works because the Slayer mythology has rules the story never breaks.

The SEA tradition offers something with that same internal completeness — a world where the spiritual and the physical are already unified, where the warrior and the mystic are the same person at different moments of the same practice, where horror and combat share a root system centuries deep.

The architecture for a story that can carry both genres simultaneously — without either cancelling the other out — already exists. It has existed for centuries.

What has been missing is the willingness to ask it the question it was never designed to answer.

That question is being built into a story now.