Context Transformer Smart Topics for Video Call Transcripts

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional video call transcription systems struggle with inaccurate identification of relevant portions, inefficient navigation, and require excessive user interactions to locate related data across multiple transcripts and content items.

Innovation Solution

A smart topic generation system utilizing a large language model, context transformer engine, and smart topic agent to automatically generate actionable smart topics from video call transcripts, summarizing and surfacing relevant content with minimal user input.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional natural language processing systems are used to generate transcripts and identify topic portions, then transcript generation is achieved, but accuracy in identifying relevant portions and related digital content deteriorates

Engineering Contradiction:

Improveaccuracy of identifying relevant portionsVSAvoidaccuracy of topic identification

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The patent introduces an intermediary system that bridges conventional NLP transcript generation and accurate topic identification. This intermediary layer processes the generated transcripts through additional analysis mechanisms to extract and verify topic relevance, thereby resolving the accuracy deterioration caused by relying solely on conventional NLP systems.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces conventional mechanical NLP-based topic identification with an enhanced system that incorporates multiple processing stages. This substitution improves accuracy by moving beyond simple keyword matching and basic NLP techniques to a more sophisticated analysis mechanism that can reliably identify relevant portions and related digital content.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Ease of operation

If conventional systems provide comprehensive transcript search functionality, then complete data access is achieved, but navigational efficiency and user interaction requirements deteriorate

Engineering Contradiction:

Improvenavigational efficiencyVSAvoidtime for user interactions

Core Design Contradiction:

Ease of operationVSLoss of time

Solution Approach 1:

The patent applies preliminary action by pre-processing and organizing transcript data into structured topic portions before user interaction is needed. This preliminary organization enables faster retrieval and more efficient navigation, reducing the time users spend searching through comprehensive transcript data while maintaining complete data accessibility.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent segments comprehensive transcripts into distinct topic portions with clear boundaries and metadata. This segmentation allows users to navigate directly to relevant sections without scrolling through entire transcripts, significantly improving navigational efficiency while preserving access to all transcript content through the segmented structure.

Inventive Principle:

Principle #1Segmentation

3Adaptability or versatility

If multiple separate interfaces and applications are used to view transcripts, aggregate data, and search content, then comprehensive functionality is achieved, but device complexity and operational simplicity deteriorate

Engineering Contradiction:

Improvecomprehensive functionalityVSAvoidnumber of interfaces and applications

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent merges multiple separate interfaces and applications into a unified system that provides comprehensive functionality. By combining transcript viewing, data aggregation, and content search capabilities into a single integrated interface, the system maintains versatile functionality while reducing the complexity of managing multiple separate applications.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent implements universality by designing a single interface that performs multiple functions: viewing transcripts, aggregating data across topics, and searching for relevant content. This multi-functional approach eliminates the need for separate applications while maintaining comprehensive functionality, thereby reducing device complexity without sacrificing adaptability.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS12581037B2Generating smart topics for video calls using a large language model and a context transformer engine

Publication Date: 2026.03.17 DROPBOX INC

AI summary

The present disclosure relates to systems, non-transitory computer-readable media, and methods for utilizing a context transformer engine, a smart topic agent, and a large language model to generate a smart topic output. In particular, in one or more embodiments, the disclosed systems generate a smart topic output from a transcript of a video call. In some embodiments, the disclosed systems provide a smart topic interface that provides the smart topic output on a client device and receives selections of smart topic elements. In one or more embodiments, the disclosed systems generate a combined smart topic from transcripts of video calls in which client devices that participated are associated with a collaborating user account group.