MOM's Magic AI

MOM's Magic AI

MOM's Magic AI

Mom’s Magic is a semi-automated AI workflow designed to digitize, organize, and activate a fragmented library of traditional Marathi recipes. By bridging the gap between raw voice data and actionable culinary insights, the system allows for real-time meal planning based on available ingredients, ensuring that the heritage of home-cooked food remains accessible even while living thousands of miles away.

Mom’s Magic is a semi-automated AI workflow designed to digitize, organize, and activate a fragmented library of traditional Marathi recipes. By bridging the gap between raw voice data and actionable culinary insights, the system allows for real-time meal planning based on available ingredients, ensuring that the heritage of home-cooked food remains accessible even while living thousands of miles away.

Mom’s Magic is a semi-automated AI workflow designed to digitize, organize, and activate a fragmented library of traditional Marathi recipes. By bridging the gap between raw voice data and actionable culinary insights, the system allows for real-time meal planning based on available ingredients, ensuring that the heritage of home-cooked food remains accessible even while living thousands of miles away.

Client

Client

Client

Prathamesh Patil

Prathamesh Patil

Prathamesh Patil

Timeline

Timeline

Timeline

45 Mins

45 Mins

45 Mins

Skills & Methods

Skills & Methods

Skills & Methods

RAG(RetrievalAugmentedGeneration)

RAG(RetrievalAugmentedGeneration)

RAG(RetrievalAugmentedGeneration)

ComputerVision

ComputerVision

ComputerVision

Problem Solving using AI

Problem Solving using AI

Problem Solving using AI

Multimodal AI

Multimodal AI

Multimodal AI

Design Operations

Design Operations

Design Operations

Workflow Automation

Workflow Automation

Workflow Automation

Google's NotebookLM

Google's NotebookLM

Google's NotebookLM

Prompt Engineering

Prompt Engineering

Prompt Engineering

User Centered Design

User Centered Design

User Centered Design

NLP (Natural Language Processing)

NLP (Natural Language Processing)

NLP (Natural Language Processing)

Team

Team

Team

Prathamesh Patil | Archana Patil (my mother) | Gemini AI | NotebookLM

Prathamesh Patil | Archana Patil (my mother) | Gemini AI | NotebookLM

Prathamesh Patil | Archana Patil (my mother) | Gemini AI | NotebookLM

PROBLEM

PROBLEM

PROBLEM

As a Design Management student living in the USA, I faced a daily cognitive load regarding meal preparation. The challenges were three-fold:

  • Fragmented Resources: My mother’s recipes were scattered across dozens of WhatsApp voice notes and text messages in Marathi. Finding a specific recipe required hours of manual playback.

  • Ingredient Misalignment: I would often find a recipe I loved, only to realize mid-way through that I lacked the necessary Indian spices or vegetables.

  • Decision Fatigue: After a long day of studio work, the mental energy required to decide what to cook, check the fridge, and cross-reference a manual archive was overwhelming. I needed a system that "knew" my pantry and my mother’s wisdom simultaneously.

As a Design Management student living in the USA, I faced a daily cognitive load regarding meal preparation. The challenges were three-fold:

  • Fragmented Resources: My mother’s recipes were scattered across dozens of WhatsApp voice notes and text messages in Marathi. Finding a specific recipe required hours of manual playback.

  • Ingredient Misalignment: I would often find a recipe I loved, only to realize mid-way through that I lacked the necessary Indian spices or vegetables.

  • Decision Fatigue: After a long day of studio work, the mental energy required to decide what to cook, check the fridge, and cross-reference a manual archive was overwhelming. I needed a system that "knew" my pantry and my mother’s wisdom simultaneously.

As a Design Management student living in the USA, I faced a daily cognitive load regarding meal preparation. The challenges were three-fold:

  • Fragmented Resources: My mother’s recipes were scattered across dozens of WhatsApp voice notes and text messages in Marathi. Finding a specific recipe required hours of manual playback.

  • Ingredient Misalignment: I would often find a recipe I loved, only to realize mid-way through that I lacked the necessary Indian spices or vegetables.

  • Decision Fatigue: After a long day of studio work, the mental energy required to decide what to cook, check the fridge, and cross-reference a manual archive was overwhelming. I needed a system that "knew" my pantry and my mother’s wisdom simultaneously.

SOLUTION

SOLUTION

SOLUTION

I developed a custom AI ecosystem that transforms raw Marathi audio into a searchable, intelligent knowledge base. The workflow leverages:

  • Multimodal Transcription: Using Gemini to transcribe and translate Marathi voice notes into structured digital recipe documents.

  • Knowledge Synthesis: Centralizing these documents into a NotebookLM RAG (Retrieval-Augmented Generation) system to act as the "brain."

  • Custom Interface (The "Gem"): A specialized Gemini Gem called "Mom's Magic" that acts as the interaction layer. It uses Computer Vision to "see" ingredients in my fridge and matches them against the RAG database to suggest the perfect recipe in real-time.

I developed a custom AI ecosystem that transforms raw Marathi audio into a searchable, intelligent knowledge base. The workflow leverages:

  • Multimodal Transcription: Using Gemini to transcribe and translate Marathi voice notes into structured digital recipe documents.

  • Knowledge Synthesis: Centralizing these documents into a NotebookLM RAG (Retrieval-Augmented Generation) system to act as the "brain."

  • Custom Interface (The "Gem"): A specialized Gemini Gem called "Mom's Magic" that acts as the interaction layer. It uses Computer Vision to "see" ingredients in my fridge and matches them against the RAG database to suggest the perfect recipe in real-time.

I developed a custom AI ecosystem that transforms raw Marathi audio into a searchable, intelligent knowledge base. The workflow leverages:

  • Multimodal Transcription: Using Gemini to transcribe and translate Marathi voice notes into structured digital recipe documents.

  • Knowledge Synthesis: Centralizing these documents into a NotebookLM RAG (Retrieval-Augmented Generation) system to act as the "brain."

  • Custom Interface (The "Gem"): A specialized Gemini Gem called "Mom's Magic" that acts as the interaction layer. It uses Computer Vision to "see" ingredients in my fridge and matches them against the RAG database to suggest the perfect recipe in real-time.

VALUE CREATED

VALUE CREATED

VALUE CREATED

Cognitive Relief: Reduced the time spent on meal-planning from 20 minutes to 30 seconds.

Zero Waste: By using Vision AI to detect ingredients, I maximized the use of fresh produce before it spoiled.

Cultural Preservation: Digitized a personal heritage of "oral recipes" that might have otherwise been lost in chat histories.

Emotional Well-being: Brought the "delightful taste of home" to my kitchen in the USA, providing a sense of comfort and heart-felt satisfaction that supported my academic focus.

Cognitive Relief: Reduced the time spent on meal-planning from 20 minutes to 30 seconds.

Zero Waste: By using Vision AI to detect ingredients, I maximized the use of fresh produce before it spoiled.

Cultural Preservation: Digitized a personal heritage of "oral recipes" that might have otherwise been lost in chat histories.

Emotional Well-being: Brought the "delightful taste of home" to my kitchen in the USA, providing a sense of comfort and heart-felt satisfaction that supported my academic focus.

Cognitive Relief: Reduced the time spent on meal-planning from 20 minutes to 30 seconds.

Zero Waste: By using Vision AI to detect ingredients, I maximized the use of fresh produce before it spoiled.

Cultural Preservation: Digitized a personal heritage of "oral recipes" that might have otherwise been lost in chat histories.

Emotional Well-being: Brought the "delightful taste of home" to my kitchen in the USA, providing a sense of comfort and heart-felt satisfaction that supported my academic focus.

STRUCTURE OF WORKFLOW

STRUCTURE OF WORKFLOW

STRUCTURE OF WORKFLOW

Capture: Mom records instructions in Marathi and sends them via WhatsApp.

  • Digitization: Audio files are processed through Gemini for high-fidelity Marathi transcription.

  • Indexing: Transcripts are fed into a NotebookLM repository, creating a structured, searchable RAG database.

  • Activation: The "Mom’s Magic" Gem serves as the UI, accepting multimodal inputs (Fridge photos or Marathi voice commands).

  • Matching: The AI identifies available ingredients and queries the recipe database to find a match.

  • Output: The system provides step-by-step instructions and "Mom’s secret tips" for the final dish.

Capture: Mom records instructions in Marathi and sends them via WhatsApp.

  • Digitization: Audio files are processed through Gemini for high-fidelity Marathi transcription.

  • Indexing: Transcripts are fed into a NotebookLM repository, creating a structured, searchable RAG database.

  • Activation: The "Mom’s Magic" Gem serves as the UI, accepting multimodal inputs (Fridge photos or Marathi voice commands).

  • Matching: The AI identifies available ingredients and queries the recipe database to find a match.

  • Output: The system provides step-by-step instructions and "Mom’s secret tips" for the final dish.

Capture: Mom records instructions in Marathi and sends them via WhatsApp.

  • Digitization: Audio files are processed through Gemini for high-fidelity Marathi transcription.

  • Indexing: Transcripts are fed into a NotebookLM repository, creating a structured, searchable RAG database.

  • Activation: The "Mom’s Magic" Gem serves as the UI, accepting multimodal inputs (Fridge photos or Marathi voice commands).

  • Matching: The AI identifies available ingredients and queries the recipe database to find a match.

  • Output: The system provides step-by-step instructions and "Mom’s secret tips" for the final dish.

SUMMARY

SUMMARY

SUMMARY

Mom’s Magic AI is a sophisticated, semi-automated workflow that preserves cultural heritage through modern AI integration. By transforming fragmented Marathi voice recordings into a structured RAG (Retrieval-Augmented Generation) database via NotebookLM, I developed a custom multimodal interface using Gemini Gems. The system utilizes Computer Vision to analyze real-time refrigerator inventory and cross-references it with a private knowledge base of traditional recipes, effectively eliminating culinary decision fatigue for busy professionals while maintaining a deep connection to home through technology.

Mom’s Magic AI is a sophisticated, semi-automated workflow that preserves cultural heritage through modern AI integration. By transforming fragmented Marathi voice recordings into a structured RAG (Retrieval-Augmented Generation) database via NotebookLM, I developed a custom multimodal interface using Gemini Gems. The system utilizes Computer Vision to analyze real-time refrigerator inventory and cross-references it with a private knowledge base of traditional recipes, effectively eliminating culinary decision fatigue for busy professionals while maintaining a deep connection to home through technology.

Mom’s Magic AI is a sophisticated, semi-automated workflow that preserves cultural heritage through modern AI integration. By transforming fragmented Marathi voice recordings into a structured RAG (Retrieval-Augmented Generation) database via NotebookLM, I developed a custom multimodal interface using Gemini Gems. The system utilizes Computer Vision to analyze real-time refrigerator inventory and cross-references it with a private knowledge base of traditional recipes, effectively eliminating culinary decision fatigue for busy professionals while maintaining a deep connection to home through technology.

© 2026 Prathamesh Patil.
© 2026 Prathamesh Patil.
© 2026 Prathamesh Patil.
© 2026 Prathamesh Patil.
© 2026 Prathamesh Patil.