MOMENTUM

Inspiration

Every creator knows this feeling: you see a trend everywhere, you want to make something around it, but you can already tell you are late.

That problem is sharper in India. Shorts trends move fast across languages, niches, creators, and communities. But most creators still discover opportunities manually by scrolling YouTube, watching competitors, and guessing what to make next.

The insight behind MOMENTUM is simple:

By the time a trend is obvious, it is usually crowded.

So I built MOMENTUM to help Indian creators catch the signal earlier.

What it does

MOMENTUM is YouTube Shorts intelligence for India.

It helps creators answer:

What is moving?
Why is it moving?
What should I create from it?

Instead of only ranking popular videos, MOMENTUM focuses on early momentum. It turns public Shorts evidence into a momentum score, a clear reason, and a creator-ready idea.

Raw Shorts -> evidence -> momentum -> explanation -> idea

The goal is simple: help creators move from scrolling and guessing to faster content decisions.

Why it matters

Creators do not just need more ideas. They need better timing.

A video with millions of views can already be saturated. A smaller Short gaining attention from a fresh niche can be the real opportunity. MOMENTUM is built around that difference:

Virality shows what already happened.
Momentum helps creators decide what to do next.

For creators, this reduces manual trend research. For agencies, it creates a repeatable discovery workflow. For brands, it surfaces emerging content opportunities before a space becomes crowded.

Why India-first

The MVP is intentionally focused:

India -> YouTube Shorts -> creator decisions

India is a high-volume, fast-moving creator market. Starting with one clear market and one platform makes the product more useful and more trustworthy.

City-level intelligence is shown as a future premium capability, but it is locked until real local data supports it. I did not want to fake local trend data just to make the product look bigger.

Product flow

Open the India-wide momentum radar.
Review Shorts signals worth attention.
Open a trend deep dive.
Check the score, source evidence, category, and niche.
Read why the signal may be moving.
Turn the signal into a creator-ready idea.
Save opportunities for later planning.

MOMENTUM does not stop at analytics. It converts analytics into action.

Architecture

MOMENTUM is built with Next.js App Router, React, TypeScript, Tailwind CSS, and server-side provider boundaries.

UI routes
  -> typed product objects
  -> server API routes
  -> demo data or YouTube adapter
  -> normalized Shorts objects
  -> deterministic scoring
  -> grounded AI interpretation

Key boundaries:

Layer	Purpose
Demo mode	Clearly labeled sample fixtures for walkthroughs
Live mode	Public YouTube signals through server-side API boundaries
Scoring	Deterministic momentum estimate before AI is used
AI	Explains evidence and creates content ideas
Trust	No fake metrics, no fake city data, no unsupported growth claims

Momentum scoring

The score is calculated before AI enters the system. MOMENTUM looks at signals such as views, recency, engagement, category, niche, and confidence around whether a video behaves like a Short.

One transparent metric is:

engagement_rate = (likes + comments) / views

Repeated snapshots can support velocity over time. A single snapshot can show a current signal, but the product does not claim historical growth unless it has the evidence.

How AI is used

AI is used as an interpretation layer, not as a fake data source.

The product first gathers evidence. Then AI helps explain:

why a signal may be moving
what pattern a creator should notice
what niche or category it belongs to
what content angle the creator can try next

AI is not allowed to invent fake metrics, city trends, guaranteed virality, or unsupported audience claims. That matters because creator tools need trust.

Features built

India-wide Shorts momentum radar
filterable trending workspace
trend deep dives with evidence and score explanation
category and niche exploration
AI-assisted creator idea generation
saved signals workspace
locked city intelligence as future premium
clear demo/live data boundaries

Challenges

The hardest challenge was balancing usefulness with honesty. It is easy to make a trend product look impressive by exaggerating the data.

I chose stricter boundaries:

demo content is labeled
live mode is separate
observed data and AI interpretation are separate
city intelligence is locked until real data supports it
momentum is a decision signal, not a guarantee of virality

What I learned

I learned that strong AI products need a clean separation between evidence, interpretation, and action.

For creators, the most useful AI is not just "generate more ideas." It is "help me choose which idea is worth acting on now."

What's next

Next steps include historical snapshots, stronger language and niche intelligence, creator-specific recommendations, city-level trend intelligence based on real data, agency exports, and feedback loops from saved ideas to creator performance.