DirecX: AI-Directed Ad Creation

teaser
App and User Flow

Inspiration

Our team lives at the intersection of creative and technical. A computer science x film double major, a professional animation artist, and a design engineer. We've sat through the late-night editing sessions. We've storyboarded shots on napkins. We understand the craft of visual storytelling because we've done it ourselves.

And we've felt the pain.

Traditional filmmaking has always been fragmented. Pre-production (planning your shots), production (capturing footage), and post-production (editing it all together). Three phases, three skill sets, often three different tools or teams. For trained filmmakers, it's manageable. For everyone else, it's a wall.

Small business owners know the ins and outs of their products better than anyone. But most don't have the time, budget, or expertise to translate that knowledge into an engaging, scroll-stopping video. Great products never get the spotlight they deserve. Not because the story isn't there, but because telling it visually is hard.

We built DirecX to collapse the entire pipeline into one app. Show Grok a photo of your product, and it doesn't just generate a generic script. It pulls real-time trending data from X to understand what's resonating right now with your target audience. Your coffee shop ad catches the local wave. Your sneaker drop rides the viral moment. Your content stays relevant because Grok is always watching the pulse of X.

Grok directs. You shoot and edit. All in one app, in your pocket.

What it does

DirecX turns the three phases of filmmaking into one seamless flow.

Step 1: Setup Your Ad Snap or upload a photo of your product. Describe your business goal, pick your ad length (15s, 30s, or 60s), and select your target audience. That's all Grok needs to get started.

Step 2: Grok Plans Your Storyboard Grok analyzes your product and generates a shot-by-shot storyboard tailored to your goals. You get shot descriptions, framing suggestions, and a teleprompter script infused with real-time X trends. No guesswork. Just follow the plan.

Step 3: Capture with Guidance Record your takes directly in the app. The teleprompter stays visible on screen so you never lose your place. Grok tells you what to shoot. You just hit record.

Step 4: Edit Your Rough Cut DirecX assembles your clips into a rough cut automatically. Trim, rearrange, or swap out shots as needed. Camera shy? Use Grok Voice to generate a professional AI voiceover instead of speaking on camera.

Step 5: Export and Share Export your polished ad and save it to your device. Ready for X, Instagram, TikTok, or wherever your audience lives.

How we built it

DirecX combines three core APIs into one unified pipeline.

Grok 4.1 Fast powers the creative brain. It analyzes product photos, generates shot-by-shot storyboards, writes teleprompter scripts, and provides real-time guidance during capture. Every decision Grok makes is informed by real-time X data, so your content stays relevant to what your audience is actually talking about.

Grok Voice API handles AI voiceovers. Users who prefer not to speak on camera can generate natural, professional narration in multiple voice styles. No recording booth required.

X API feeds trending data directly into the script generation process. This ensures ads feel timely and contextual rather than generic.

On the development side, we built the iOS app in Swift/SwiftUI. AVFoundation handles video capture, playback, and the editing pipeline. The teleprompter and shot guidance overlays run on Core Animation for smooth, real-time performance during recording. We used WebSocket connections for streaming Grok Voice audio and stitching AI voiceovers into the final export.

The result is a single app where storyboard, capture, and edit all happen in one place. No switching tools. No fragmented workflow. Product in, ad out.

Challenges we ran into

The hackathon track disappeared. Our team originally planned to compete in the Grok Movie track. We have members who studied film and animation, and we had a solid storyboard-driven concept ready to go. Then, right before the hackathon started, the competition track was removed. Surprise.

Pivoting under pressure. We had to rethink our entire approach overnight. The core storyboard idea survived, but we restructured everything around ad creation for the Grok Ads Studio track. It forced us to scope ruthlessly and focus on what actually mattered.

Real-time teleprompter sync. Coordinating Grok's script output with the recording UI required careful state management. The teleprompter needed to scroll smoothly while the camera was rolling without blocking the capture pipeline.

Video editing in 24 hours. Building a functional trim, rearrange, and replace editor from scratch under hackathon time pressure pushed our AVFoundation knowledge to the limit.

Grok Voice integration. Handling WebSocket audio streaming and cleanly stitching AI voiceovers into the final video export involved more late-night debugging than we expected.

Accomplishments that we're proud of

We shipped a complete pipeline. Storyboard generation, guided capture, editing, voiceover, and export. All working. All in 24 hours.

We pivoted and still delivered. The track change could have derailed us. Instead, we adapted and came out with something more focused and practical than our original concept.

We integrated three xAI/X APIs. Grok API, Grok Voice API, and X API all working together in a single cohesive workflow. Not just demos. Actual integration.

We built something we'd use ourselves. That's the bar. DirecX isn't a hackathon gimmick. It's a tool we genuinely want to keep developing.

What we learned

Grok Voice latency is workable. Streaming audio over WebSocket and generating voiceovers in real time is surprisingly smooth once you get the pipeline right.

X trending data is powerful. Injecting real-time context into scripts makes the output feel alive. Generic scripts feel stale by comparison.

Constraints sharpen the product. Losing the Movie track forced us to cut scope and focus. The result is tighter than our original plan would have been.

Film knowledge transfers. Our team's background in film and animation helped us think in shots, pacing, and visual storytelling. That foundation made the storyboard generation logic feel natural rather than forced.

What's next for DirecX

DirecX is just getting started. The next evolution is live streaming.

Imagine going live to promote a product launch. AR overlays highlight features in real time. Text and filters react to what you're saying. Viewers engage directly with the stream, not just passively watching.

A key feature we're building is real-time reply analysis. Grok monitors live chat sentiment as you stream, helping you gauge audience interest and adjust your delivery, pacing, or talking points on the fly. No more guessing if your content is landing.

By combining AI direction, live streaming, AR overlays, and real-time analytics, DirecX will empower creators and businesses to host interactive product launches, promotional events, and campaigns that feel personal and immediate.

We're not just making ads easier to create. We're redefining how ads are experienced on X.