/Patent Tracker
GooglePatent applicationAI & softwareInterest 6/10

Google patents way to remember images in long AI chats

This filing describes a method for AI systems to remember images shared during a conversation. When a user sends a picture, the system converts it into a simplified digital code and saves it. If the user mentions that image later in the chat, the system pulls the saved code back up so the AI can still understand what is being discussed.

Why it mattersThis helps AI assistants keep track of visual details over long conversations, preventing users from having to re-upload or re-describe images they already shared.

In the filing’s own words

Implementations relate to handling visual content across a multi-turn dialog. A user input that includes natural language content and visual content is received during the dialog. If the visual content is being received for the first time in the dialog, the visual content is processed to generate a corresponding tokenized representation of the visual content. The corresponding tokenized representation can be cached in a database in association with the dialog, or in association with a user account of a user of the user query. If the visual content is subsequently referenced in the dialog, the corresponding tokenized representation of the visual content is retrieved from the database. The corresponding tokenized representation of the visual content, corresponding tokenized representations of natural language content, and optionally other metadata can be processed, using a generative model, to generate a response responsive to the user input.

What exactly they claim (claim 1)

1. A method implemented by one or more processors, the method comprising: maintaining a structured file including a sequence of dialog entries for a multi-turn human-to-computer dialog, the structured file including at least a first entry representing previously processed visual content, and the first entry including an explicit visual content marker identifying a tokenized representation of the visual content; receiving user input during the multi-turn human-to-computer dialog; in response to receiving the user input, identifying the explicit visual content marker within the sequence of dialog entries in the structured file; retrieving, from the structured file and based on the identified explicit visual content marker, the tokenized representation of the visual content; and generating, using a generative model, a response to the user input based on processing at least the tokenized representation of the visual content.

Facts

Company
Google (GOOGLE LLC)
Official title
PRESERVATION OF VISUAL CONTENT ACROSS MULTI-TURN DIALOGS WITH GENERATIVE MODEL(S)
Publication number
US20260289126A1
Application number
19686030
Filed
May 22, 2026
Published
Sep 24, 2026
Inventors
7
Claims
20

Classified under

G06F 40/284 Physics & computingG06V 20/70 Physics & computing
Read the full filing (official PDF) ↗ Open in the Patent Tracker

More from Google in AI & software

Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.