/Patent Tracker
AmazonGranted patentAI & softwareInterest 6/10

Amazon patents system that helps AI assistants remember specific items

This patent describes a method for managing specific items, or entities, mentioned in user requests to a language model. The system stores detailed information about these items in a temporary memory area and uses unique codes to retrieve that data when the AI needs it to answer a question.

Why it mattersThis technique could make virtual assistants more accurate by ensuring they have the correct details about specific products or people when processing complex requests.

In the filing’s own words

Devices and techniques are generally described for entity management. In various examples, a natural language request comprising a reference to an entity may be received. Resolved entity data associated with the entity may be determined. Identifier data associated with the resolved entity data may be generated and the resolved entity data may be stored in association with the identifier data in a session cache. A first prompt that comprises the identifier data may be generated. A language model (LM) may generate a data request that includes the identifier data as an input parameter. At least a portion of the resolved entity data may be received from the session cache using the data request. A second prompt that comprises at least the portion of the resolved entity data may be generated and used to perform an action with respect to the entity in response to the natural language request.

What exactly they claim (claim 1)

1. A computer-implemented method comprising: receiving a first natural language request; determining a first reference to a first entity in the first natural language request; executing an application programming interface (API) call comprising the first entity to an entity resolution service; determining, based on the API call, first resolved entity data comprising structured metadata indicating a plurality of attributes of the first entity; generating first identifier data for the first resolved entity data; storing, in a first memory location of a session cache, the first resolved entity data; generating a mapping between the first identifier data and the first memory location of the session cache in which the first resolved entity data is stored; generating a first language model (LM) prompt based on the first natural language request, wherein the first LM prompt comprises the first identifier data; determining, by an LM using the first LM prompt, that additional information about the first entity is used to provide a response to the first natural language request; generating, by the LM, a data request applicable to the first natural language request, wherein the data request comprises an API call comprising the first identifier data as an input parameter; receiving, in response to the API call, at least a portion of the structured metadata of the first resolved entity data from the session cache; generating a second LM prompt comprising at least the portion of the structured metadata; generating, by the LM using the second LM prompt, a first natural language response to the first natural language request, wherein the first natural language response provides information about the first entity according to at least the portion of the structured metadata; and outputting the first natural language response.

Facts

Company
Amazon (Amazon Technologies, Inc.)
Official title
Entity management for language models
Patent number
US12738277B1
Application number
18541595
Filed
Dec 15, 2023
Granted
Sep 15, 2026
Inventors
6
Claims
20

Classified under

G10L 15/285 Physics & computingG10L 15/22 Physics & computingG10L 2015/223 Physics & computing

More from Amazon 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.