Inspiration

Influencer marketing in India is booming, but for MSMEs and D2C founders it's still a empty box — how do you know if a creator's followers are real, what a "fair" rate even looks like, or whether your money is safe once you pay. I built CollabKar to fix that, using Synepto AI to bring transparency, trust, and speed to brand-creator collaborations.

What it does

CollabKar is an AI-powered influencer marketplace that lets brands describe their campaign in plain English and instantly get matched with verified creators. Synepto AI handles three things automatically: fake-follower detection (99.2% accuracy), fair-price rate benchmarking in INR, and milestone-secured escrow payments — so brands and creators can collaborate without the usual guesswork or risk.

How I built it

I built CollabKar with a natural-language search layer powered by Synepto AI, which breaks down a brand's description and maps it to creator niches, audience quality , and rate benchmarks. The platform tracks engagement patterns and follower growth to flag bot activity, and wraps payments in an escrow flow that releases funds on campaign milestones rather than upfront.

Challenges I ran into

Getting the fake-follower detection to be reliable without flagging genuinely fast-growing creators was tricky — sudden spikes aren't always fraud. I also had to balance making the AI matching feel conversational and simple while still surfacing enough data (rates, engagement, niche) for brands to trust the recommendation.

Accomplishments that we're proud of

I'm proud of getting natural-language campaign matching working end-to-end. I also built out a full escrow-secured payment flow within a 24 hours hackathon.

What's next for Collabkar

Next, I want to expand Synepto AI's matching to support multi-platform campaigns (Instagram, YouTube, and beyond), add automated campaign performance reporting for brands, and grow our vetted creator network beyond the initial 2,000+ creators. and also turn it into a startup.

Built With

docker
express.js
framer-motion
javascript
next.js
node.js
python
react
tailwindcss
vercel

Submitted to

All Things Agentic Hackathon

Created by

I built CollabKar end-to-end for this hackathon — frontend, backend, and the AI integration.

On the frontend, I built the brand and creator dashboards using Next.js and Tailwind, deployed on Vercel. On the backend, I set up the Express API with MongoDB and Redis, deployed on Render via Docker.

The core challenge was the AI layer — using Gemini to power natural-language creator matching and fair-price rate suggestions. Getting the model to return consistent, structured pricing data (rather than free-form text) took a fair amount of prompt iteration. I also built the fake-follower detection logic, which was tricky to tune so it didn't flag creators who were just genuinely growing fast.

This was my first time integrating Gemini directly into a production-style app, and I learned a lot about designing API contracts around an LLM's non-deterministic output.

KennethAnsel Kenneth