Obsyr

AI that can learn and automate any software

Sahil Singhavi, Founder & CEO

01

The Problem

40% of time wasted on tasks that could be automated

Humans waste countless hours performing repetitive tasks on computers
Enterprise CTOs estimate $13.2M+ wasted annually on tasks that could be automated
Current solutions break with UI changes or hallucinate actions
Knowledge workers struggle to automate cross-application workflows

02

Our Solution

Obsyr is an AI system that learns any software through observation and then executes tasks based on user prompts.

Learns by Watching

Builds cognitive models of applications by observing human interactions, understand the context and purpose behind actions

Adapts to Changes

Maintains adaptability when interfaces change, unlike brittle RPA tools

Continuously Improves

Gets smarter with every execution through reinforcement learning

03

Product Demo

See Obsyr in action, learning and executing tasks across any software.

Observation Mode

Watches and learns from your actions

Execution Mode

Performs tasks based on natural language prompts

Error Recovery

Adapts to unexpected situations and UI changes

04

Business Model

Obsyr's business model combines cutting-edge AI technology with a tiered SaaS approach to deliver exceptional value to knowledge workers and enterprises.

Key Partners

Enterprise software vendors
Cloud infrastructure providers
AI research institutions
System integrators
IT consultancies

Key Activities

AI model development & training
Software observation engine
Task execution framework
User experience optimization
Enterprise integration

Value Propositions

Learn any software through observation
Execute tasks via natural language
76% task completion accuracy
Adapts to UI changes automatically
Saves 11+ hours weekly per user
Works across multiple applications
Continuous learning & improvement

Customer Relationships

Self-service for individuals
Dedicated support for teams
Enterprise account management
Community knowledge sharing
Continuous feedback loops

Customer Segments

Knowledge workers
Creative professionals
Software developers
Enterprise IT departments
Business operations teams
Users of Legacy or Specialized Software

Key Resources

Proprietary AI models
Software interaction dataset
Computer vision technology
Technical talent
Secure cloud infrastructure

Channels

Direct website sales
Enterprise sales team
Partner network
App marketplaces
Developer community

Cost Structure

AI research & development
Cloud computing resources
Engineering team
Sales & marketing
Enterprise customer support

Revenue Streams

Individual tier: $79/month
Team tier: $149/user/month
Enterprise tier: $299/user/month
Custom enterprise solutions
API access for developers

05

Market Opportunity

1.1B Knowledge workers globally

$200B+ Productivity & automation market

$320B Total addressable market by 2030

The global business productivity software market is projected to grow by $119.4 billion from 2025-2029. By capturing just 1% of knowledge workers at our Individual tier pricing, Obsyr would generate $10.4B in annual revenue.

Key Market Segments (2025)

Enterprise IT Automation: $50.2B

Knowledge Worker Productivity: $70.4B

Creative Professional Tools: $13.9B

Software Development Automation: $59.9B

06

Our Technology

AI Models

Custom fine-tuned vision-language models (ViT-L/14)

Multi-agent system with specialized AI agents

Graph neural networks for UI relationship modeling

Transformer-based attention for context understanding

Reinforcement learning for error recovery (70% success)

Core Technology Stack

Java core framework for system-level interactions

Python for AI components and computer vision

PyTorch for custom computer vision models

AWS for computing, hosting model and infrastructure

React for web interface and user interaction

Docker for containerization and deployment

Data & Training

1,200+ hours of software interaction training data

Each additional hour improves performance by ~0.4%

Custom dataset of UI elements across 150+ common applications

Multi-stage detection pipeline with 83% precision

Performance Metrics

UI Element Detection

82%

Task Completion

76%

Error Recovery

70%

Multi-Task Learning

85%

07

Competition

Unlike traditional automation tools that break with UI changes or LLM agents that hallucinate actions, Obsyr builds a cognitive model of applications through observation.

Obsyr

RPA Tools

UiPath

Automation Anywhere

Blue Prism

Power Automate

AI Agents

Adept AI

OpenAI

Anthropic

UI-TARS

Task Completion

76%

High for unchanged UI

38%

Adapts to UI Changes

Setup Complexity

Low

High

Medium

Cross-App Workflows

Continuous Learning

UI Element Detection

82%

Selector-based

~65%

Error Recovery

70%

Limited

Variable

Learning from Examples

85% after 3 examples

Multiple attempts

Failure Rate with UI Changes

24%

65-70%

~50%

Context Understanding

Execution Speed

Fast

Very Fast

Slow

08

Traction

21

Active Beta Users

87%

Weekly Retention

17

Tasks Automated Weekly

11.2

Hours Saved Weekly

Application Usage Breakdown

Microsoft Office (42%)

Adobe Creative Suite (31%)

Web Applications (27%)

Roadmap

Now

21 active beta users with 87% retention

3 Weeks

Expand beta to 50 users

3-4 Months

Full product launch

09

Team

Sahil Singhavi

Founder & CEO

Education: BTech Computer Science (JK Lakshmipat University), Machine Learning Specialization Online (Stanford University)

FACO Care: Founded health-tech startup at 19, developed AI for respiratory disease detection with 70% accuracy, secured $100K funding, deployed in 3 health centers during covid near me

Kaufy Tech: Developed AI-powered conversion optimization platform increasing e-commerce conversion by 27%, secured 3 enterprise clients generating $15K MRR

Vybex: Created AI-driven fashion recommendation platform with 85% prediction accuracy, achieved 109 paying customers ($2,180 MRR) within 6 months

TaskNinja: Co-founded task management platform handling concurrent requests, built ML-based matching algorithm improving completion rates by 40%, exited for $72K

Jyot: Led development of assistive technology for visually impaired, built computer vision system with 65% accuracy, improved accessibility for 500+ users

Project Kai: Co-founded blockchain/AI venture, created "Nikki" AI-powered 3D influencer with 7K+ followers, implemented custom rendering pipeline, exited for $94K

Freelance: Completed 75+ projects with 98% satisfaction rate, specialized in AI, automation, and design projects for healthcare and mental health applications

Open Source: Contributed to TensorFlow's audio processing libraries and PyTorch's computer vision modules, improving processing speed by 5%

Investment & Recognition

Personally invested $55K to bootstrap Obsyr's development, with runway to expand our technology.

Awards & Competitions

Covieshield Hackathon 2020 - First Place ($100K)

UC Berkeley Hackathon 2020 - Winner ($15K AWS credits)

IIT Bombay Techfest Coding Competition - Team Winner

IIT Delhi Roborace 2020 - Runner-up

State Science Olympiad - Finalist (High School)

Technical Expertise

AI & Computer Vision: TensorFlow, PyTorch, OpenCV

Full-Stack Development: React, Node.js, MongoDB

3D & Graphics: Unity, Custom Rendering Pipelines

Mobile: Flutter, React Native, Native iOS/Android

Certifications

Google UX Design Professional (2023)

Google Augmented Reality Specialization (2022)

Technion Startup Entrepreneurship (2021)

Stanford Machine Learning Specialization (2021)

University of Sydney Critical Thinking (2020)

freeCodeCamp Full Stack Web Development (2020)

10

Our Vision

A world where AI handles most computer tasks, freeing humans to focus on creativity and innovation.

Autonomous Digital Assistant

Proactive task suggestion based on user behavior

Collaborative Intelligence

Shared task libraries across organizations

General Purpose Automation

Expansion beyond UI-based tasks to API and system-level automation

11

Thank You

Obsyr is transforming human-computer interaction by creating an AI system that can learn, use, and master any software through observation and practice.

singhavisahil@gmail.com

obsyr.com

Demo Access

Website: demo.obsyr.com

Username: demo

Password: #DemoTime

12