Forem: Vishal Narayan

I Kept Quitting Habit Apps — So I Built One That Punishes Me If I Do.

Vishal Narayan — Thu, 09 Apr 2026 13:27:39 +0000

Most habit apps failed me.

Not because they were bad… but because I was inconsistent.

I’d start strong, track everything perfectly for a few days, then miss one day — and that was enough to spiral into quitting entirely.

No real consequences. No accountability. Just another broken streak.

So I decided to experiment with a different idea:

What if failing had a cost?

💡 The Core Idea

Instead of just tracking habits, I built something that adds commitment.

The app is called Covenant.

Here’s how it works:

You stake credits before starting a habit
You complete your task (gym, coding, reading, etc.)
You submit proof (AI-verified photo or synced health/app data)
If you succeed → you get your credits back
If you fail → they’re burned

No second chances. No “I’ll make it up tomorrow”.

🧠 Why This Approach?

Most habit systems rely on motivation.

But motivation is unreliable.

I wanted to explore three things instead:

Skin in the Game

When there’s something at risk, your behavior changes. Even a small stake creates pressure to follow through.

Accountability

Doing things alone is easy to abandon. Adding squad-based visibility makes inconsistency visible.

Friction Against Cheating

Most apps trust self-reporting. I didn’t.

So I added:

AI-based photo verification
Health data syncing
App usage tracking

The goal was to reduce “fake progress” without making the experience painful.

⚙️ How I Built It
Frontend: React Native (Expo)
Backend: Supabase (Auth, Postgres, Realtime)
AI Verification: Groq (Llama Vision) + Gemini
Styling: NativeWind

One of the hardest parts wasn’t building features — it was balancing strictness vs usability.

Too strict → people quit
Too loose → system becomes meaningless

⚠️ Challenges I Ran Into
Designing a system that feels fair even when users fail
Preventing obvious ways to game verification
Avoiding “gambling” mechanics while still having stakes
Keeping onboarding simple despite a complex idea
🚀 What I Learned
People don’t need better reminders — they need better systems
Social pressure is underrated in habit building
Adding consequences changes behavior more than adding rewards
📱 If You Want to Try It

https://play.google.com/store/apps/details?id=com.keepcovenant.app

🤔 Open Questions

I’m still figuring a lot of this out:

Does this build long-term discipline or just short-term compliance?
Is losing your stake motivating… or discouraging?
How much verification is too much?

Would love to hear your thoughts — especially if you’ve tried building or using habit systems before.

I Kept Quitting Habit Apps — So I Built One That Punishes Me If I Do

Vishal Narayan — Thu, 09 Apr 2026 13:26:39 +0000

Most habit apps failed me.

Not because they were bad… but because I was inconsistent.

I’d start strong, track everything perfectly for a few days, then miss one day — and that was enough to spiral into quitting entirely.

No real consequences. No accountability. Just another broken streak.

So I decided to experiment with a different idea:

What if failing had a cost?

💡 The Core Idea

Instead of just tracking habits, I built something that adds commitment.

The app is called Covenant.

Here’s how it works:

No second chances. No “I’ll make it up tomorrow”.

🧠 Why This Approach?

Most habit systems rely on motivation.

But motivation is unreliable.

I wanted to explore three things instead:

Skin in the Game

When there’s something at risk, your behavior changes. Even a small stake creates pressure to follow through.

Accountability

Doing things alone is easy to abandon. Adding squad-based visibility makes inconsistency visible.

Friction Against Cheating

Most apps trust self-reporting. I didn’t.

So I added:

AI-based photo verification
Health data syncing
App usage tracking

The goal was to reduce “fake progress” without making the experience painful.

⚙️ How I Built It
Frontend: React Native (Expo)
Backend: Supabase (Auth, Postgres, Realtime)
AI Verification: Groq (Llama Vision) + Gemini
Styling: NativeWind

One of the hardest parts wasn’t building features — it was balancing strictness vs usability.

Too strict → people quit
Too loose → system becomes meaningless

https://play.google.com/store/apps/details?id=com.keepcovenant.app

✨ Just shipped a fun project – PopChoice, an AI-powered movie recommendation system 🎥

Vishal Narayan — Tue, 02 Sep 2025 11:47:51 +0000

I just finished building a project called PopChoice 🎬 – it’s an AI-powered movie recommendation app that suggests films based on your preferences.

🔧 Tech highlights:

Nomic embeddings + Supabase vector DB for semantic search

LLM-driven recommendations via Groq

Enriched with OMDb API posters & metadata

Frontend: React + Vite with a modern glassmorphic UI

👉 Try it live: pppppop-choice-app.netlify.app

👉 Code: GitHub Repo

Would love your feedback on:

UX / UI polish

Recommendation quality

Features you think would make it better

Thanks in advance!

Personal Dashboard — new-tab extension with dynamic backgrounds, weather, quick links & daily facts

Vishal Narayan — Sun, 31 Aug 2025 13:15:13 +0000

I built a new-tab browser extension called Personal Dashboard that replaces your new tab with a customizable dashboard: dynamic Unsplash backgrounds (category selection + smart text-color), real-time time/date with personalized greetings, location-based weather, customizable quick links with favicon support, and daily advice + interesting facts.

Install as an unpacked extension in Chrome/Edge/Brave (or load temporary in Firefox). API keys are required — .env.example and setup steps are in the repo.

Uses Unsplash, OpenWeatherMap, AdviceSlip and API-Ninjas for content; preferences are stored in localStorage.

Built with vanilla JavaScript (ES6 modules), CSS Grid/Flexbox and glassmorphism UI. MIT licensed — feedback and PRs welcome.

Repo: https://github.com/vishalnarayan2809/personal-dashboard

🤖 I built a Chrome extension that makes my resume less embarrassing using AI (and it actually works?)

Vishal Narayan — Thu, 14 Aug 2025 17:05:32 +0000

So I was tired of copy-pasting the same generic resume for every job and getting crickets back. Like, why would a startup care that I'm "proficient in Microsoft Office" when they're looking for someone who can "move fast and break things"?

Enter AI Presume - a Chrome extension that takes your boring resume and job description, throws them at Groq's Llama model, and spits out a tailored resume that actually makes sense for the role.

The weird part? It works entirely in your browser. No sketchy servers, no "we'll store your data safely" promises. Your resume stays on your machine while the AI does its magic.

Even weirder: It exports to PDF with actual selectable text and clickable links. I know, revolutionary stuff in 2025.

Built it because I was procrastinating on actual job applications, but now I'm actually using it for... actual job applications. The irony is not lost on me.

Some features that make me unreasonably proud:

Keeps your formatting intact (finally, an AI that respects my carefully crafted margins)
One-page limit with auto-compression (because nobody reads page 2 anyway)
Works on all the Chromium browsers your company forces you to use
Privacy-first (your embarrassing employment gaps stay between you and the AI)

Fair warning: It's Manifest V3, so Firefox users get the "it might work, it might not" experience. Chrome users get the full treatment.

Check it out if you're also tired of explaining why your 2019 "WordPress blog maintenance" experience makes you qualified for a Senior DevOps role: https://github.com/vishalnarayan2809/AI-Resume-extension

P.S. - Yes, I used this extension to tailor my resume for posting about this extension. Meta level: achieved.

TL;DR: Made a thing that makes job hunting slightly less soul-crushing. It's free, it's local, and it might help you get past those ATS robots that definitely judge your Times New

I proved ChatGPT wrong for the first time!

Vishal Narayan — Sat, 26 Jul 2025 05:40:52 +0000