# US 2026/0212405 A1 · Method and system for generative artificial intelligence based assistant for a specialty retail store

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0212405 A1
- Document ID: US20260212405A1
- Status: Published application
- Kind: A1
- Application number: 19/338,596
- Filed: 2025-09-24
- Published: 2026-07-23
- Earliest claimed: 2025-01-23
- Assignees: Tata Consultancy Services Limited
- Applicants: Tata Consultancy Services Limited
- Inventors: JEISOBERS THIRUNAVUKKARASU; SELVA GANAPATHY ULAGANATHAN; VAIRA MUTHU THANGAVEL; HAREESH KUMAR; DHANASEKARAN GOPAL; SUTIRTHA SAHU
- Web page: https://patentgrep.com/patent/US20260212405A1

## CPC

- G06Q 30/02017: {Market opportunity analysis, e.g. gap analysis}
- G06Q 30/0639: {Locating goods or services, e.g. based on physical position of the goods or services within a shopping facility}
- G06N 3/045: Combinations of networks
- G06N 20/00: Machine learning
- G06Q 30/0205: {based on location or geographical consideration}
- G06Q 30/06313: {based on similarity of goods or services, e.g. substitute or alternate goods or services}
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

In a competitive retail world, store specific competitive strategies need to be formed by understanding unique characteristics of assortment followed by each specialty competitor within a trade area. Retailers face challenges to get intensive knowledge about store specific competitor assortment due to practical difficulties. Embodiments herein provide a method and system for a generative artificial intelligence (GenAI) based assistant to implement one or more competitive retail strategies associated with specialization of a specialty retail store. Locating competitive position of a specialty retail store and identifying gap in assortment is achieved by capturing unique characteristics of a specialty competitor store which is enabled by applying dynamic enrichment of content in which direction for the enrichment provided by approximating existing patterns that occur among the specialty competitive stores.

## Claims

### Claim 1

A processor-implemented method comprising:
receiving, via an Input/Output (I/O) interface, location of a retail specialty store of a retailer;
determining, via one or more hardware processors, a plurality of unknown stores within a predefined trade area of the retail specialty store using a web mapping platform, wherein the predefined trade area depends on a category of the retail specialty store;
identifying, via one or more hardware processors, an e-commerce website of each of the plurality of unknown stores and the retailer specialty store using a web crawling technique;
obtaining, via one or more hardware processors, one or more details of an e-commerce website of each of the plurality of unknown stores and the retailer specialty store, wherein the one or more details include an information associated with one or more items, a price, an image, and a video of each of the one or more items and a visionary statement explaining core value of the plurality of unknown stores and the retailer specialty store;
storing, via the one or more hardware processors, the obtained one or more details of the ecommerce website of each of plurality of unknown stores and the retailer specialty store into an individual specific containers separately in an unstructured format;
enhancing, via the one or more hardware processors, the content of each container by applying iteratively at least one contextual embedding technique, wherein the at least one contextual technique includes a Bidirectional Encoder Representations from Transformers (BERT), and Named-entity recognition (NER);
computing, via the one or more hardware processors, a dynamic cosine similarity for any two containers with the dynamic enriched content and repeating to all pairwise containers to derive a dynamic cosine similarity matrix;
identifying, via the one or more hardware processors, one or more latent variables, cumulative variance contribution and one or more factor loadings by applying a principal component analysis (PCA) on the computed dynamic cosine similarity matrix;
maximizing, via the one or more hardware processors, the identified cumulative variance contribution to select an ideal enriched content from each container of the plurality of unknown stores and the retailer specialty store and store into an associated enriched retrieval augmented generation (RAG) file; and
performing, via the one or more hardware processors, a plurality of tasks associated with specialty retailing using the selected ideal enriched content of one store among the unknown stores and the retailer specialty store, wherein the plurality of tasks includes (a) getting response with improved relevance to GenAI queries related with one store among the plurality unknown stores and the retailer specialty store, and (b) locating gap in online assortment for the retailer specialty store.

### Claim 2 (depends on 1)

The processor-implemented method of claim 1, comprising:
analyzing, via the one or more hardware processors, the identified one or more latent variables, and one or more the factor loadings to derive intention of the plurality of unknown stores within the predefined trade area of the location of the retail specialty store;
identifying, via the one or more hardware processors, at least one competitive store and level of competition by determining position of the plurality of unknown stores within latent variable based on the specialization type, magnitude of the one or more factor loadings, wherein the magnitude of factor loading and distinct specialization type are considered to place the plurality of unknown stores in the top position of the competitive stores list;
observing, via the one or more hardware processors, a difference between the identified at least one competitive store and the specialty retail store by applying the multimodal GenAI models on each category of corresponding the first set of containers and the second set of containers with enriched contents; and
generating, via the one or more hardware processors, one or more recommendations based on the observed difference between the identified at least one competitive store and the specialty retail store using the enriched contents, wherein the generated one or more recommendations are displayed to an end user for executing one or more corrective actions.

### Claim 3 (depends on 1)

The processor-implemented method of claim 1, wherein the plurality of tasks comprising:
getting response with improved relevance to GenAI queries related with one store among the plurality unknown stores and the retailer specialty store by applying GenAI in which the improved relevancy is measured in comparison to the response received from text without enrichment; and
locating gap in online assortment for the retailer specialty store using the enriched RAG files with enriched content and to identify the position of the retail specialty store with respect to specialization in the trade area.

### Claim 4 (depends on 1)

The processor-implemented method of claim 1, wherein the one or more latent variables represent the type of specialization of the plurality of stores including unknown stores and retailer store in which the plurality of stores having specific type of items form a latent variable.

### Claim 5 (depends on 1)

The processor-implemented method of claim 1, wherein the latent variables with the specialty retailer store is identified and the plurality of unknown stores in the identified latent variable are the competitive stores and level of competition is identified from magnitude of factor loading of each unknown stores within the identified latent variable.

### Claim 6 (depends on 1)

The processor-implemented method of claim 1, wherein the maximizing of cumulative variance contribution enables:
enriching the content of container in two phases such as (i) in initial phase, enrichment happens within the scope defined as prompts in prompt engineering and to move towards natural grouping that exist among the specialty stores and (ii) in later phase, enrichment happens automatically towards hyper specialization that exist naturally among the stores;
maximizing depth of differentiation among the stores with respect to specialization; and
enriching the content by the direction of existing groups among the stores in the form of specialization.

### Claim 7

A system comprising:
a memory storing instructions;
one or more Input/Output (I/O) interfaces; and
one or more hardware processors coupled to the memory via the one or more I/O interfaces, wherein the one or more hardware processors are configured by the instructions to:
receive location of a retail specialty store of a retailer;
determine a plurality of unknown stores within a predefined trade area of the retail specialty store using a web mapping platform, wherein the predefined trade area depends on a category of the retail specialty store;
identify an e-commerce website of each of the plurality of unknown stores and the retailer specialty store using a web crawling technique;
obtain one or more details of an e-commerce website of each of the plurality of unknown stores and the retailer specialty store, wherein the one or more details include an information associated with one or more items, a price, an image, and a video of each of the one or more items and a visionary statement explaining core value of the plurality of unknown stores and the retailer specialty store;
store the obtained one or more details of the ecommerce website of each of plurality of unknown stores and the retailer specialty store into an individual specific containers separately in an unstructured format;
enhance the content of each container by applying iteratively at least one contextual embedding technique, wherein the at least one contextual technique includes a Bidirectional Encoder Representations from Transformers (BERT), and Named-entity recognition (NER);
compute a dynamic cosine similarity for any two containers with the dynamic enriched content and repeating to all pairwise containers to derive a dynamic cosine similarity matrix;
identify one or more latent variables, cumulative variance contribution and one or more factor loadings by applying a principal component analysis (PCA) on the computed dynamic cosine similarity matrix;
maximize the identified cumulative variance contribution to select an ideal enriched content from each container of the plurality of unknown stores and the retailer specialty store and store into an associated enriched retrieval augmented generation (RAG) file; and
perform a plurality of tasks associated with specialty retailing using the selected ideal enriched content of one store among the unknown stores and the retailer specialty store, wherein the plurality of tasks includes (a) getting response with improved relevance to GenAI queries related with one store among the plurality unknown stores and the retailer specialty store, and (b) locating gap in online assortment for the retailer specialty store.

### Claim 8 (depends on 7)

The system of claim 7, wherein the one or more hardware processors are configured by the instructions to:
analyze the identified one or more latent variables, and one or more the factor loadings to derive intention of the plurality of unknown stores within the predefined trade area of the location of the retail specialty store;
identify at least one competitive store and level of competition by determining position of the plurality of unknown stores within latent variable based on the specialization type, magnitude of the one or more factor loadings, wherein the magnitude of factor loading and distinct specialization type are considered to place the plurality of unknown stores in the top position of the competitive stores list;
observe a difference between the identified at least one competitive store and the specialty retail store by applying the multimodal GenAI models on each category of corresponding the first set of containers and the second set of containers with enriched contents; and
generate one or more recommendations based on the observed difference between the identified at least one competitive store and the specialty retail store using the enriched contents, wherein the generated one or more recommendations are displayed to an end user for executing one or more corrective actions.

### Claim 9 (depends on 7)

The system of claim 7, wherein the plurality of tasks comprising:
getting response with improved relevance to GenAI queries related with one store among the unknown stores and the retailer specialty store by applying GenAI in which the improved relevancy is measured in comparison to the response received from text without enrichment; and
locating gap in online assortment for the retailer store using the RAG files with enriched content and to identify the position of the retail specialty store with respect to specialization in the trade area.

### Claim 10 (depends on 7)

The system of claim 7, wherein the one or more latent variables represent the type of specialization of the plurality of stores including unknown stores and retailer store in which the plurality of stores having specific type of items form a latent variable.

### Claim 11 (depends on 7)

The system of claim 7, wherein the latent variables with the specialty retailer store is identified and the plurality of unknown stores in the identified latent variable are the competitive stores and level of competition is identified from magnitude of factor loading of each unknown stores within the identified latent variable.

### Claim 12 (depends on 7)

The system of claim 7, wherein the maximizing of cumulative variance contribution enables:
enriching the content of container in two phases such as (i) in initial phase, enrichment happens within the scope defined as prompts in prompt engineering and to move towards natural grouping that exist among the specialty stores and (ii) in later phase, enrichment happens automatically towards hyper specialization that exist naturally among the stores;
maximizing depth of differentiation among the stores with respect to specialization; and
enriching the content by the direction of existing groups among the stores in the form of specialization.

### Claim 13

One or more non-transitory machine-readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause:
receiving, via an Input/Output (I/O) interface, location of a retail specialty store of a retailer;
determining a plurality of unknown stores within a predefined trade area of the retail specialty store using a web mapping platform, wherein the predefined trade area depends on a category of the retail specialty store;
identifying an e-commerce website of each of the plurality of unknown stores and the retailer specialty store using a web crawling technique;
obtaining one or more details of an e-commerce website of each of the plurality of unknown stores and the retailer specialty store, wherein the one or more details include an information associated with one or more items, a price, an image, and a video of each of the one or more items and a visionary statement explaining core value of the plurality of unknown stores and the retailer specialty store;
storing the obtained one or more details of the ecommerce website of each of plurality of unknown stores and the retailer specialty store into an individual specific containers separately in an unstructured format;
enhancing the content of each container by applying iteratively at least one contextual embedding technique, wherein the at least one contextual technique includes a Bidirectional Encoder Representations from Transformers (BERT), and Named-entity recognition (NER);
computing a dynamic cosine similarity for any two containers with the dynamic enriched content and repeating to all pairwise containers to derive a dynamic cosine similarity matrix;
identifying one or more latent variables, cumulative variance contribution and one or more factor loadings by applying a principal component analysis (PCA) on the computed dynamic cosine similarity matrix;
maximizing the identified cumulative variance contribution to select an ideal enriched content from each container of the plurality of unknown stores and the retailer specialty store and store into an associated enriched retrieval augmented generation (RAG) file; and
performing a plurality of tasks associated with specialty retailing using the selected ideal enriched content of one store among the unknown stores and the retailer specialty store, wherein the plurality of tasks includes (a) getting response with improved relevance to GenAI queries related with one store among the plurality unknown stores and the retailer specialty store, and (b) locating gap in online assortment for the retailer specialty store.

### Claim 14 (depends on 13)

The one or more non-transitory machine-readable information storage mediums of claim 13, wherein the one or more instructions which when executed by the one or more hardware processors further cause:
analyzing the identified one or more latent variables, and one or more the factor loadings to derive intention of the plurality of unknown stores within the predefined trade area of the location of the retail specialty store;
identifying at least one competitive store and level of competition by determining position of the plurality of unknown stores within latent variable based on the specialization type, magnitude of the one or more factor loadings, wherein the magnitude of factor loading and distinct specialization type are considered to place the plurality of unknown stores in the top position of the competitive stores list;
observing a difference between the identified at least one competitive store and the specialty retail store by applying the multimodal GenAI models on each category of corresponding the first set of containers and the second set of containers with enriched contents; and
generating one or more recommendations based on the observed difference between the identified at least one competitive store and the specialty retail store using the enriched contents, wherein the generated one or more recommendations are displayed to an end user for executing one or more corrective actions.

### Claim 15 (depends on 13)

The one or more non-transitory machine-readable information storage mediums of claim 13, wherein the one or more instructions which when executed by the one or more hardware processors further cause:
getting response with improved relevance to GenAI queries related with one store among the plurality unknown stores and the retailer specialty store by applying GenAI in which the improved relevancy is measured in comparison to the response received from text without enrichment; and
locating gap in online assortment for the retailer specialty store using the enriched RAG files with enriched content and to identify the position of the retail specialty store with respect to specialization in the trade area.

### Claim 16 (depends on 13)

The one or more non-transitory machine-readable information storage mediums of claim 13, wherein the one or more latent variables represent the type of specialization of the plurality of stores including unknown stores and retailer store in which the plurality of stores having specific type of items form a latent variable.

### Claim 17 (depends on 13)

The one or more non-transitory machine-readable information storage mediums of claim 13, wherein the latent variables with the specialty retailer store is identified and the plurality of unknown stores in the identified latent variable are the competitive stores and level of competition is identified from magnitude of factor loading of each unknown stores within the identified latent variable.

### Claim 18 (depends on 13)

The one or more non-transitory machine-readable information storage mediums of claim 13, wherein the maximizing of cumulative variance contribution enables:
enriching the content of container in two phases such as (i) in initial phase, enrichment happens within the scope defined as prompts in prompt engineering and to move towards natural grouping that exist among the specialty stores and (ii) in later phase, enrichment happens automatically towards hyper specialization that exist naturally among the stores;
maximizing depth of differentiation among the stores with respect to specialization; and
enriching the content by the direction of existing groups among the stores in the form of specialization.
