The USA Patent Workplace not too long ago revealed Google’s continuation on a patent for a search system that detects when there is no such thing as a passable reply for a question and waits to mechanically ship the reply when it turns into accessible.
Search And AI Assistant
The patent, revealed in February 2026, is a continuation of an older patent, with the principle modifications being to use this patent throughout the context of an AI assistant. The invention describes fixing the issue of answering a query when no precise reply is on the market on the time a person makes the question. What it does is waits till there’s a passable reply, at which level it circles again to the person with the reply, with out them having to ask once more.
The patent is titled, Autonomously offering search outcomes post-facto, together with in assistant context. Though the patent mentions high quality thresholds, these thresholds are outlined within the sense of whether or not the reply meets the person’s wants.
The patent describes six situations that will set off the invention:
- When no search outcomes meet outlined high quality or authoritative-answer standards.
- When outcomes exist however fail to offer a definitive or authoritative reply that satisfies these standards.
- When no outcomes meet high quality standards as a result of the data is just not but accessible.
- When a question seeks a selected reply and no end result satisfies the required standards.
- When a useful resource later satisfies the outlined standards after beforehand missing required data.
- When a beforehand accessible useful resource is refined or up to date in order that it now meets the standards.
Helpful And Full Solutions
Google’s patent says that the invention is an answer for occasions when there is no such thing as a helpful or full solutions as a result of the data doesn’t but exist or is just not adequate, forcing customers to maintain looking repeatedly.
The system checks if outcomes meet:
- A top quality commonplace
- Authoritativeness commonplace
- Or a completeness commonplace.
If the present solutions don’t meet these requirements, the system will retailer the question and monitor for brand new or up to date data. As soon as it turns into accessible it’s going to ship the outcomes to the person later with out them looking once more.
Comply with-Up Questions Are Not Obligatory
What’s novel in regards to the invention is that it permits follow-up supply of outcomes after the unique question with out requiring a brand new follow-up questions. It additionally surfaces search outcomes proactively in notifications or assistant conversations.
At a later time, when new or up to date data turns into accessible that satisfies the standards, the system proactively delivers that data to the person. This supply can happen via notifications, inside an unrelated interplay, or throughout a later dialog with an automatic assistant.
The system can also optionally notify the person that no good outcomes are at the moment accessible and ask in the event that they wish to be told when higher outcomes seem.
What this method does is it transforms search from a one-time, user-initiated motion right into a persistent, ongoing course of the place the system continues working within the background and updates the person when significant data turns into accessible.
Cross-Machine Continuity
An fascinating characteristic of this invention is that it could attain out to the person throughout a number of gadgets.
Right here is the place it’s outlined:
[0012] In some implementations, the question is acquired on a further computing machine that’s along with the computing machine for which the content material is offered for presentation to the person.”
This capabiilty is highlighed once more in part [0067]:
“For instance, the content material could also be offered for presentation to the person through the identical computing machine the person utilized to submit the question and/or through a separate computing machine.”
It may possibly additionally go cross-device as a visible and/or audible output throughout gadgets and within the type of an automatic assistant, and may current the data when the person is interacting with the automated assistant in a unique context, describing an “ecosystem” of gadgets.
Lastly, the patent explains that the data may be surfaced when the person is interfacing with the automated assistant in a totally completely different context:
[0040]”…the content material could also be offered for presentation to the person through the identical computing machine the person utilized to submit the question and/or through a separate computing machine. The content material could also be offered for presentation in varied varieties. For instance, the content material could also be offered as a visible and/or audible push notification on a cellular computing machine of the person, and could also be surfaced impartial of the person once more submitting the question and/or one other question.
Additionally, for instance, the content material could also be introduced as visible and/or audible output of an automatic assistant throughout a dialog session between the person and the automated assistant, the place the dialog session is unrelated to the question and/or one other question searching for related data.”
Takeaways
The patent (Autonomously offering search outcomes post-facto, together with in assistant context) is consistent with Google’s imaginative and prescient of tasked-based agentic search, the place AI assistants assist customers accomplish issues. This patent may very well be utilized to an AI agent that’s requested for tickets to an occasion when the tickets aren’t but accessible. Or it may very well be utilized to creating restaurant reservations when the reservations when the dates open up. Each of these situations are associated to task-based agentic search (TBAS)
Listed below are seven takeaways:
- The system shops knowledge related to the person about unresolved queries, permitting it to trace unanswered data wants over time moderately than treating every search as a one-off occasion.
- It delivers outcomes inside future interactions, together with unrelated assistant conversations, not simply via standalone notifications.
- The notifications can occur throughout an ecosystem of gadgets.
- An absence of outcomes is outlined by failing to satisfy high quality standards, which may be the absence of knowledge, the reply not being accessible but, or the reply is just not accessible from authoritative sources.
- The system focuses on queries that search particular solutions, moderately than basic informational searches.
- It helps cross-device continuity, enabling a question on one machine to be fulfilled in a while one other.
- The design reduces repeated searches by eliminating the necessity for customers to verify again, then autonomously circling again when the data is on the market.
Featured Picture by Shutterstock/uyabdami