This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle
What I Built
Creators can spend hours preparing sponsored videos, only to discover that their script contains the wrong discount code, a missing hashtag, or a claim the sponsor prohibited. These mistakes cause unnecessary revisions and delay approvals.
SponsorProof compares a brand's campaign brief with a creator's script or caption to identify potential issues before submission.
It is designed for:
-Independent content creators managing sponsorships.
-Freelance video editors reviewing sponsored content.
-Small marketing agencies handling creator campaigns.
The project checks campaign requirements such as promotional codes, hashtags, sponsorship disclosures, sponsor links, prohibited claims, competitor mentions, and required talking points.
A potential business model would be a free version for individual creators and a paid team version for agencies that need shared briefs, reports, and collaboration features.
Demo
Live website: https://sponsorproof.onrender.com
The website includes an example NordShield VPN campaign. Its brief requires the promo code CREATOR70, while the example script contains CREATOR20. The audit workflow is designed to identify this mismatch and other campaign requirements that need attention.
The hosted website currently supports deterministic audit mode. Gemma-powered analysis runs through my local Ollama setup; the hosted Ollama model connection is not yet configured. I am sharing this limitation rather than claiming that hosted AI analysis works.
Code
GitHub repository: https://github.com/architsinghal0005/sponsorproof
The repository contains the application source code and project documentation.
How I Built It
I built SponsorProof using:
-Next.js for the web application.
-TypeScript for application logic and type safety.
-Tailwind CSS for the user interface.
-Gemma 3 1B through Ollama for local AI inference.
-Rule-based validation for exact campaign requirements.
-Render to deploy the public web application.
One key design decision was to combine deterministic checks with AI. Exact requirements, such as promotional codes and hashtags, are better handled with predictable rules. A language model can help interpret more nuanced instructions, but its findings still need verification.
Running Gemma locally also gave me a way to experiment with open-weight AI without depending entirely on a paid hosted model API. Local processing can keep campaign material on the user's machine when the application is configured not to send that material elsewhere.
The hosted deployment and local AI setup are separate at this stage, and connecting the hosted application to a reliable model endpoint remains future work.
Why Does Open Innovation Matter?
Open innovation makes it possible for individual developers to experiment with AI and build useful tools without depending entirely on closed, proprietary services.
Using an open-weight model such as Gemma gives developers more flexibility over where inference runs, how prompts are designed, and how a model is adapted to a specific workflow.
For SponsorProof, this matters because campaign briefs and unreleased scripts can contain sensitive business information. Local inference offers an opportunity to process that material locally, provided the application does not transmit it elsewhere.
Open models also make experimentation more accessible: developers can test different approaches, inspect failure cases, and combine AI with conventional software engineering techniques.
There are trade-offs. Local inference requires suitable hardware, smaller models can miss context, and AI-generated findings should not be treated as legal or advertising-compliance certification.
SponsorProof is a prototype exploring how open-weight AI and deterministic software checks can work together to reduce repetitive review work.
Prize Categories
-Best Use of Gemma
-Best Use of Render
Top comments (0)