I Built India's First AI Vedic Astrology Platform in 17 Days — Here's Everything I Did

From zero to Page 1 on Google, 500+ users, and a trademark filed — as a solo founder.

The Idea

India has a ₹40,000 crore astrology market. AstroTalk makes ₹600 crore/year. AstroSage has been around for 20 years.

But nobody had built an AI-powered, free, Hindi-first Vedic astrology platform.

So I built one.

In 17 days.

The Stack

• 
  Framework: Next.js 14 (App Router)


• 
  Language: TypeScript


• 
  Styling: Tailwind CSS + inline styles


• 
  AI: Groq API (llama-3.3-70b-versatile) — for kundli reading generation


• 
  Astrology Calculations: AstrologyAPI.com (Swiss Ephemeris under the hood)


• 
  Deployment: Vercel


• 
  Email: Nodemailer + Zoho SMTP


• 
  Domain: kundliai.in

Total infra cost per month: under ₹500

What I Built in 17 Days

The Technical Challenges

Challenge 1: Groq Rate Limits

The biggest headache. Groq's free tier has strict rate limits and when traffic spikes, llama-3.3-70b-versatile hits the ceiling fast.

My fix — model fallback chain:

const models = [

  'llama-3.3-70b-versatile',

  'llama-3.1-8b-instant',

  'gemma2-9b-it',

];



for (const model of models) {

  try {

    const groqRes = await fetch('https://api.groq.com/openai/v1/chat/completions', {

      method: 'POST',

      headers: { 'Authorization': `Bearer ${process.env.GROQ_API_KEY}` },

      body: JSON.stringify({ model, messages, max_tokens: 1000 }),

    });

    const data = await groqRes.json();

    if (!data.error) {

      reading = data.choices[0].message.content;

      break;

    }

    await new Promise(r => setTimeout(r, 1000));

  } catch { continue; }

}

If all models fail, show a graceful fallback with the chart data instead of a broken error.

Challenge 2: AstrologyAPI Field Names

AstrologyAPI.com has some quirks. The nakshatra field is literally spelled Naksahtra (typo in their API that they never fixed):

const nakshatra = lagnaData?.Naksahtra || lagnaData?.nakshatra || '—';

Took me an embarrassing amount of time to figure that out.

Challenge 3: Hindi AI Reading

Getting Groq to respond in proper Devanagari Hindi consistently required very specific prompting:

const langInstruction = lang === 'hi'

  ? 'Respond entirely in Hindi (Devanagari script). Do not use any English words.'

  : 'Respond in English with occasional Sanskrit terms for authenticity.';

Challenge 4: Windows Double Extension Bug

Every file I downloaded got saved as page.tsx.tsx. Had to add this to my workflow permanently:

mv "page.tsx.tsx" "page.tsx"

SEO Strategy — From 0 to Page 1 in 17 Days

This is what I'm most proud of.

Day 1-17 content output:

• 20 blogs on the website (Hindi + English mix)


• 16 posts on Medium


• 45 Quora answers


• Crunchbase, LinkedIn, Wellfound, SlideShare profiles


• Trademark filed on Day 14

Result after 17 days:

• Average position: 7.6 (Page 1)


• Monthly impressions: 531
  


• CTR: 6% (industry average is 2-3%)


• Users: 500+
  

The key insight: Hindi content has 10x less competition than English for the same keywords.

aaj ka rashifal — 823K monthly searches, difficulty 42

kundali — 450K monthly searches, difficulty 64

rahu kaal today — 201K monthly searches, difficulty 42

Most competitors only target English. I went Hindi-first from Day 1.

The Prompt Engineering

The core of KundliAI is the kundli reading prompt. Here's the architecture:

1. Calculate real planetary positions via AstrologyAPI (Swiss Ephemeris)

2. Build structured astrology context string

3. Feed to Groq with persona prompt

4. Return personalized 400-500 word reading

The persona matters enormously:

You are Pandit JyotishAI, a master Vedic astrologer 

with 30 years of experience. Use the EXACT planetary 

positions given. Mention their actual Lagna, Moon sign, 

Nakshatra by name. Be warm, wise and make them feel 

truly seen by the cosmos.

Generic AI responses feel cold. Adding the persona + real chart data = responses that feel genuinely personal.

Revenue Plan

Currently at ₹0 — Razorpay integration coming this week.

Planned monetization:

• Kundli PDF Report — ₹99


• Kundli Matching Report — ₹149


• Numerology Report — ₹99


• Love Compatibility Report — ₹99


• Life Timeline Report — ₹199


• Premium Subscription — ₹199/month

Target: ₹1L/month by Month 6.

What's Next

• Razorpay payment integration


• Google Sheets email list capture


• City landing pages (Delhi, Mumbai, Bangalore)


• SEObot for automated blog generation


• Predis.ai for Instagram automation

Lessons for Solo Founders

1. Ship on Day 1, improve forever

I launched with basic features and added 11 more over 17 days. Waiting for perfection = waiting forever.

2. Vernacular is an unfair advantage

Hindi content = less competition, more trust, better CTR. Most devs ignore this.

3. Free tools can power a real product

Groq free tier + AstrologyAPI + Vercel hobby = production-ready platform at ₹0.

4. SEO compounds

Every blog, every Quora answer, every backlink compounds. Start Day 1, not Month 3.

5. Trademark early

Filed on Day 14. Costs ₹4,500 in India. Worth every rupee.

Try It

👉 kundliai.in — Free AI Vedic Astrology in Hindi & English

Generate your complete Vedic birth chart in 30 seconds — Lagna, Moon sign, Nakshatra, Mahadasha, planetary positions — all free.

Building in public. Follow along for weekly updates on traffic, revenue and product.

If you're building something in the Indian market — especially targeting Hindi users — happy to share more about the SEO strategy that got us to Page 1 in 17 days.