Free AI micro drama workflow with Claude and Google Flow

The Rise of Micro Dramas and a Free Production Workflow

0:00 to 0:29 in the video, under a minute

Vertical micro dramas have become massive drivers of engagement on social platforms, often accumulating millions of views. However, most available production tutorials rely on costly subscription tools and paid models such as Seedance 2.5, which quickly drain user credits.

To solve this problem, a streamlined end-to-end framework makes it possible to script, generate, and edit multi-scene vertical dramas entirely with free AI platforms. The system relies on Claude for scripting and prompt engineering, Google Flow for consistent media generation, and CapCut for final post-production.

Developing the Genre, Concept, and Timed Script

0:30 to 1:51 in the video, about 1 minute

The production starts by attaching a structured master workflow document into a large language model such as Claude or ChatGPT. Claude outlines standard vertical drama genres, such as billionaire romance, revenge stories, and secret identities. Once a genre is selected, the model produces ten high-hook storyline concepts to choose from.

After selecting a premise, feeding Claude specialized micro drama scriptwriting instructions yields a comprehensive screenplay package. This output includes scene-by-scene timing, dialogue limited to fit vertical attention spans, camera shot descriptions, continuity notes, and an audited beat sheet to ensure proper dramatic pacing.

• Select from trending vertical narrative tropes like workplace romance or secret heirs.
• Generate a structured 90-second timeline broken into 10-second scene increments.
• Maintain control over character dialogue, scene transitions, and narrative twists.

Designing Consistent Characters and Voices in Google Flow

1:52 to 4:00 in the video, about 2 minutes

Maintaining character consistency across multiple cuts is usually the biggest challenge in AI video creation. Claude handles this by generating detailed character descriptions alongside tailor-made image prompts specifying age, wardrobe, styling, and demeanor.

Inside Google Flow, users navigate to the Characters tab and paste these prompts to establish primary reference portraits. The platform allows creators to build multi-angle body sheets featuring front, back, and close-up views. Assigning a dedicated synthetic voice profile to each character at this stage guarantees consistent vocal identity across every line of dialogue.

Automating Video Generation with Flow Agent

4:01 to 5:34 in the video, about 2 minutes

Rather than generating scenes one by one, Claude compiles a dedicated text file containing granular, cinematic prompts for every individual shot. Each entry details the aspect ratio, lighting style, environment, active characters, physical movements, and exact lines spoken.

In Google Flow, project settings are configured to a vertical 9:16 aspect ratio using the Omni Flash model. After feeding text-to-video instructions to the platform's autonomous agent, creators supply the full prompt file alongside the stored character references. The agent executes the entire prompt sequence in order, systematically delivering ready-to-use clips for the full story.

• Configure video generation default settings to 9:16 vertical orientation.
• Provide system instructions to the Flow agent to force sequential shot generation.
• Approve required generation credits once to automate the entire batch queue.

Assembling, Refining, and Captioning in CapCut

5:35 to 6:00 in the video, under a minute

Once all scene generations are completed, each clip must be reviewed individually. Any shots with glaring motion artifacts or consistency mistakes can be regenerated before upscaling and downloading. Renaming exported files by their scene number prevents confusion during post-production.

The clips are then imported into video editing software like CapCut and placed sequentially onto the timeline. Editors can trim micro-glitches, insert subtle dissolve transitions between cuts, and generate automated dynamic captions to ensure high retention on mobile feeds.

Final Thoughts

Producing cinematic vertical micro dramas no longer requires expensive enterprise AI software or extensive manual asset management. By chaining structured prompts in Claude with automated agent generation in Google Flow, creators can rapidly turn an initial premise into a polished 90-second narrative.

With automated character sheets, fixed voice casting, and simple assembly in tools like CapCut, this free pipeline allows anyone to publish high-volume, episodic storytelling tailored for modern mobile audiences.