Inspiration

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for VIDSLOOM

Inspiration

Small businesses need short-form video to win attention, but most owners do not have the time, team, or production rhythm to keep up with fast-moving social platforms. VIDSLOOM was built for that gap: a practical AI-operated video growth engine that turns a business offer, proof, target audience, brand voice, and approval rules into campaign assets that can be reviewed, posted, measured, and improved.

What it does

VIDSLOOM helps businesses produce short-form marketing campaigns without asking the owner to become a video editor. A customer provides the business context: what they sell, who they serve, proof points, brand voice, assets, target platforms, preferred video length, quality settings, notification rules, and social-account readiness. VIDSLOOM then creates trend-aware campaign concepts, hooks, scripts, captions, thumbnails, CTAs, video samples, posting windows, approval queues, and manual or direct-posting workflows.

The product is built around approval-first operations. Direct social posting is permission-gated: customers do not give VIDSLOOM social passwords, and auto-posting only starts after a customer connects the relevant platform account and approves publishing rules. Where direct posting is not available, VIDSLOOM still gives the customer a posting-ready kit.

How we built it

VIDSLOOM is a Next.js and TypeScript application deployed on Google Cloud Run. Production storage uses Firestore and Cloud Storage, async work is routed through Cloud Tasks and Cloud Scheduler, and operational evidence is exposed through health checks, publishing queues, media-generation jobs, and ops-alert endpoints. The AI layer uses Gemini through Google Cloud to support trend intelligence, campaign planning, scripts, creative direction, quality gates, and media generation workflows.

The product also includes SendGrid-backed follow-up, a VIDSLOOM-native newsletter/autoresponder system, Stripe Checkout, customer onboarding, customer asset intake, OAuth-aware social posting, YouTube Shorts upload support, and visual QA scripts for the public landing videos and pages.

AI-native operations

VIDSLOOM is not only an app that calls AI. The business workflow itself is operated through AI-assisted systems:

  • Offer and audience intake turns messy owner inputs into structured campaign briefs.
  • Trend intelligence maps current short-form patterns to the customer's category.
  • Script and creative agents produce hooks, scripts, captions, shot lists, thumbnails, CTAs, and schedule windows.
  • Quality gates check proof usage, claims, safe-zone readability, duration, and customer approval state.
  • Publishing automation decides whether each post is ready, blocked, scheduled, manually postable, or waiting for account permissions.
  • Follow-up automation notifies the owner and keeps opted-in prospects in a warm-market newsletter sequence.
  • Evidence workflows capture live posts, screenshots, views, enquiries, bookings, sales notes, and next-test decisions.

Humans still handle customer relationships, final business judgment, compliance decisions, and final approval. AI handles the repeatable work that normally prevents small businesses from producing consistent video marketing.

Challenges we ran into

The hard part was making VIDSLOOM feel like a real business system instead of a demo generator. Short-form video marketing has many failure modes: fake-looking visuals, unreadable overlays, hallucinated proof, unsupported claims, unsafe social posting, weak customer follow-up, and unclear revenue evidence. We solved this with a hybrid pipeline: Gemini plans and generates creative assets, while deterministic rendering and QA systems keep exact text, proof, claims, logos, CTAs, and posting state controlled.

Social auto-posting was another constraint. Direct posting requires customer OAuth and platform approvals. VIDSLOOM therefore ships both direct posting where available and manual posting kits everywhere else, so the customer still gets a hands-off production workflow without unsafe credential handling.

Accomplishments

  • Production site deployed at https://vidsloom.com.
  • Cloud Run production service deployed in business-heroes-infinity.
  • Customer-facing landing page, pricing, audit form, pilot flow, newsletter, checkout, policy pages, and mobile-first UI.
  • Internal campaign workspace with trend intelligence, campaign generation, video assets, approval queue, publishing queue, calendar, evidence capture, lead follow-up, newsletter operations, and ops alerts.
  • AI media-generation pipeline with duration selection, quality mode, prompt compilation, reference-frame/video/TTS adapters, media QA, and final assembly.
  • Public sample videos with automated and AI-assisted visual QA.
  • YouTube Shorts OAuth/upload path implemented for approved customer accounts.
  • Stripe checkout and SendGrid follow-up configured.

What we learned

Small businesses do not want another content tool. They want reliable business outcomes: more enquiries, bookings, appointments, product sales, warm follow-up, and clear proof of what was posted. VIDSLOOM therefore needs to behave like a video growth operator: collect the right business inputs, generate the campaign, manage approvals, post or prepare the post, notify the customer, and record evidence for the next cycle.

What's next

Before the final deadline, VIDSLOOM should replace the current OAuth-focused demo video with a stronger full XPRIZE demo showing the production workflow end to end: lead capture, customer intake, Gemini trend intelligence, generated videos, approval queue, posting kit/direct posting, evidence capture, follow-up automation, Stripe checkout, and revenue/customer proof. The final submission should also update verified arms-length revenue, expenses, customer evidence, and corporate ID if available.

Built With

Share this project:

Updates