Studio

An AI-powered ad agency that turns your product into a viral X campaign and learns from every post

Comment

Inspiration

We find it very hard to market our projects. Despite the various communities on X, gaining a huge traction for products we're building is a near impossible task, especially with a lack of funds and the time commitment and strategy it takes to create viral accounts. We decided to build a product ourselves, one that we would use in the future, and small advertisers can use as well.

What it does

Given a product URL and a budget, it develops a full ad campaign with multiple posts, AI-generated images and videos, predicted CTR and conversions, and when to post. One click to generate media, and it schedules a timing to post to X automatically. The analytics side pulls your own post history, shows what's working, and feeds those insights back into future campaigns. There's a pipeline builder where you can drag your analytics through AI filters and wire it directly into campaign creation. The more you use it, the better it gets at knowing what works for you and how you can improve your ads.

How we built it

We used Next.js 14 with TypeScript. Tailwind and shadcn/ui for the UI. Grok handles strategy generation and images. Google Veo handles videos. X posting uses OAuth 1.0a. The pipeline builder is custom React with SVG lines for connections. Data flows through nodes and can pass through LLM filters. Storage is JSON files.

Challenges we ran into

We had a hard time to identify a solution that could solve our problem because we wanted to generate quality content for an affordable manner. We found it very hard when Cursor or other AI agents would use legacy APIs or wrong API methods for X in such a way where we had to scour documentation ourselves in a short time frame.

Accomplishments that we're proud of

We're proud to have completed an end-to-end product that can drive a meaningful impact in today's advertisement space. You can go from a product URL to a scheduled X post with generated media in minutes. The pipeline builder works and it was built from scratch and is personalized based on what data the advertiser wants and insights. The predictions show real reasoning/rationale, not just random numbers, and the UI is beautiful if I say so myself. And the analytics can actually feed back into future campaigns through the user, so it gets better over time.

What we learned

We learned different perspectives of how X treats AI ads, especially the fact that X doesn't want AI slop in everyone's feeds for ads. Despite, we knew that by combining data pipelines, analytics, and quality models such as Grok with its reasoning capabilities, we could break through this barrier.

What's next for Studio

We hope to expand Studio to every small advertiser who'd like to make a meaningful impact on their ad campaign for a much cheaper cost than what's currently on the market for ad agencies.

Built With

Share this project:

Updates