Our Portfolio

AI Products
Built to Scale

7+ production-ready AI solutions serving 5M+ users

Astro AI

AI-powered Vedic astrology platform delivering personalized insights via WhatsApp with 200% engagement boost

The Challenge

Traditional astrology consultations are expensive ($50-$200 per session), time-consuming, and inaccessible to most people. Users face high costs, limited availability, complexity barriers, and app fatigue. 70% of 18-35 year olds are interested in astrology but only 15% have had professional readings due to cost and access barriers.

Our Solution

Built an AI-powered Vedic astrology platform combining Swiss Ephemeris for NASA-grade astronomical calculations with GPT-4 for natural language conversations, delivered via WhatsApp. The system provides personalized birth chart analysis, daily predictions, compatibility matching, and multilingual support (English, Hindi, Gujarati) - all for $2.99/month vs $50-200 per traditional consultation.

OpenAI GPT-4
WhatsApp Business API
LLM Fine-tuning
Swiss Ephemeris
Python
PostgreSQL
RAG Pipeline

Key Features

WhatsApp Integration
Seamless delivery via WhatsApp Business API, no app download required
Vedic AI Engine
Custom LLM fine-tuned on authentic Vedic astrology texts
Birth Chart Generation
Complete Kundli with planetary positions, houses, nakshatras using Swiss Ephemeris
Personalized Predictions
Daily, weekly, monthly forecasts based on real-time transits
200%
Engagement Boost
5M+
Target Users
4.8/5
User Satisfaction
3s
Avg Response Time

MBTI Matching

Comprehensive personality matching algorithm integrating MBTI framework with dating preferences for transparent, science-based matchmaking

The Challenge

Dating app had stagnant engagement despite 100K user base. Average session time stuck at 3 minutes, low message rate (2.3 per match vs industry 5+), 65% first-week churn, and no product differentiation. Users complained matches felt random with no transparency in how the algorithm works, making it hard to start meaningful conversations.

Our Solution

Designed and implemented ground-up personality matching system combining MBTI (Myers-Briggs Type Indicator) psychological framework with existing preferences. Created transparent 0-100% compatibility scores with detailed explanations across 4 dimensions: Personality Compatibility (40%), Value Alignment (30%), Lifestyle Compatibility (20%), and Communication Style (10%). System provides educational insights about personality dynamics.

Python
PostgreSQL
MBTI Framework
Algorithm Design
A/B Testing

Key Features

16 MBTI Personality Types
Comprehensive assessment covering all personality dimensions
Transparent Compatibility Scores
0-100% scores with detailed breakdowns explaining why you match
Psychology-Based Algorithm
Weighted scoring based on complementary traits and cognitive functions
Educational Insights
Learn about personality compatibility and relationship dynamics
200%
Session Time Increase
152%
Messages per Match
133%
Match Acceptance
100K+
Users Served

Voice UXR Agent

AI-powered voice agent conducting, transcribing, and analyzing user research interviews at scale with 70% cost reduction

The Challenge

Product team needed 10x more user feedback but manual interviews were limiting: only 2-3 interviews/day capacity, $50/interview cost, 3 days from interview to insights, and couldn't scale without proportional cost increase. Attempted solutions like hiring more researchers (too expensive) and survey tools (low 12% response rates) failed.

Our Solution

Built AI-powered voice agent system automating entire interview pipeline: outbound calling (Twilio), natural voice conversations (ElevenLabs + GPT-4), real-time transcription (Whisper), sentiment analysis, theme extraction, and insights generation. Scaled capacity from 2-3 to 50+ interviews daily while reducing cost from $50 to $15 per interview.

ElevenLabs
OpenAI GPT-4
Whisper API
Twilio
Python
Retool
PostgreSQL

Key Features

Natural Voice Conversations
Human-like voice with empathetic responses and graceful interruption handling
Dynamic Follow-Up Questions
GPT-4 generates contextual follow-ups based on user responses
Real-Time Transcription
95% accuracy with speaker diarization and timestamps
Sentiment Analysis
Per-response sentiment scoring detecting frustration, excitement, confusion
70%
Cost Reduction
100+
Daily Interviews
25x
Capacity Increase
$7K+
Monthly Savings

Marketing Dashboard

Unified analytics dashboard consolidating 5+ data sources with AI-powered sentiment analysis, saving $10,000+ monthly and 15+ hours weekly

The Challenge

Marketing team drowning in fragmented tools spending $1,200/month on Hootsuite ($300), Sprout Social ($500), and Mention ($400). Team spent 15+ hours/week copying data between tools, 2 days to generate weekly reports, with no sentiment analysis or unified view. Slow reaction to trending content and missed viral opportunities.

Our Solution

Built all-in-one analytics dashboard using Retool consolidating Instagram, Twitter, TikTok, PostgreSQL, and Google Analytics. Replaced 3 expensive SaaS tools with custom solution providing better insights. Added AI-powered sentiment analysis, viral content pattern identification, competitor tracking, and automated daily insight summaries—all in one interface.

Retool
OpenAI GPT-4
Instagram API
Twitter API
TikTok API
PostgreSQL
Google Analytics 4
Python

Key Features

Unified Data Integration
Real-time data from Instagram, Twitter, TikTok, PostgreSQL, Google Analytics
AI-Powered Sentiment Analysis
Analyzes all comments with -1 to +1 scoring, flags negative comments
Viral Content Analysis
Pattern identification for what makes content go viral, replication playbooks
Competitor Tracking
Monitor 3 main competitors with engagement comparison and gap analysis
$10K+
Monthly Savings
15h
Weekly Time Saved
25%
Content Improvement
534%
Year 1 ROI

AI Chatbot

Personality-aware conversation AI fine-tuned on 450+ examples, generating personalized icebreakers and increasing session time by 100%

The Challenge

35% of matches never received a first message. Generic openers like 'Hey' got only 15% response rate. Average conversation length was just 3.2 messages before dying. Users reported not knowing what to say. 42% churned after no successful conversations in first week. Pre-written templates felt impersonal (8% usage) and random questions weren't contextual (12% response rate).

Our Solution

Fine-tuned GPT-3.5 on 450+ high-quality conversation examples to create personality-aware chatbot generating AI-powered conversation starters based on users' MBTI types. System generates 3-5 personalized icebreakers, contextual follow-up suggestions, and 'keep the conversation going' prompts—all tailored to personality compatibility dynamics between matches.

OpenAI GPT-3.5
LLM Fine-tuning
MBTI Framework
Python
PostgreSQL
React

Key Features

Personality-Aware Starters
3-5 personalized conversation openers based on MBTI compatibility
Contextual Follow-Ups
Dynamic suggestions during conversation when chat momentum slows
Training on 450+ Examples
High-quality conversations from dating coaches and successful users
MBTI Integration
Understands all 16 personality types and their communication dynamics
100%
Session Time Increase
50K+
Daily Conversations
78%
Response Rate
100K+
Users Served

Poll Engine

Automated campaign management interface reducing launch time from 2 days to 5 minutes with 85% manual effort reduction

The Challenge

Manual campaign creation took 2 days involving spreadsheets, SQL queries, and cross-team coordination. Required engineering team for each launch. Dependent on expensive Amazon Personalize ($800/month) with limited control. No self-serve interface for content team. Errors in manual targeting led to poor campaign performance and team bottlenecks.

Our Solution

Built self-serve campaign management platform with visual campaign builder (no code/SQL required), precision audience targeting (age, gender, location, personality type, behavior), preview before launch, A/B test setup, real-time performance tracking, and automated scheduling. Replaced Amazon Personalize with custom targeting algorithm, eliminating $800/month dependency.

React
TypeScript
Redux
Node.js
Express
PostgreSQL
Redis
Bull
Socket.io
AWS

Key Features

Visual Campaign Builder
No-code interface for creating targeted poll campaigns
Precision Audience Targeting
Filter by age, gender, location, MBTI type, user behavior
Campaign Preview
Estimate reach and view sample audience before launch
A/B Test Setup
Split audiences and compare performance automatically
99%
Launch Time Reduction
85%
Manual Effort Saved
35%
Campaign Performance
3x
More Campaigns/Month

Web Onboarding

Web-based MBTI assessment and app conversion funnel generating $1,500/month revenue (30% boost) from first month

The Challenge

All personality features locked behind app download barrier. Users wanted to try before downloading. No web presence for SEO/discoverability. Missing monetization of 10K+ monthly web visitors. No email lead capture before app install. App store approval barrier created download hesitation with no preview of product value.

Our Solution

Built web-based MBTI personality assessment with monetization funnel: SEO-optimized landing page → 16-question test with beautiful UI → Personality results page with free basic insights → Premium $4.99 detailed PDF report OR seamless app download with data transfer → Email nurture sequence. Created entirely new revenue stream generating $1,500 in first month.

React
TypeScript
Stripe
Node.js
PostgreSQL
Email Marketing
PDF Generation
Google Analytics

Key Features

SEO-Optimized Landing Page
'Discover Your Dating Personality Type' targeting 50K+ monthly searches
Mobile-Optimized Assessment
16-question test with progress bar, 5-minute completion
Instant Results
Personality type reveal with celebration animation and detailed description
Premium PDF Reports
15-page comprehensive analysis for $4.99 (40% conversion)
$1.5K
Monthly Revenue (Mo. 1)
30%
Revenue Boost
10K+
Monthly Web Visitors
40%
Premium Conversion