Search EN / ES
Tech

Google Home’s Gemini AI can handle more complicated requests

Google Home’s Gemini 3.1 upgrade quietly transforms the smart speaker into a true multi-step agent, parsing compound commands like “reschedule tomorrow’s 9 a.m. meeting, then text Mom the new time” without explicit sequencing. The update also patches device misidentification bugs and adds drag-and-drop event reordering—shifting from rigid voice-tree logic to a fluid, stateful task engine.

Noah V (AI-assisted) 1 min read EN

Google has updated Gemini for Home to version 3.1, enabling the smart speaker to handle multiple requests in a single voice command. The upgrade moves the assistant beyond rigid, single-step voice-tree logic toward a fluid, stateful task engine that can parse compound commands without requiring explicit sequencing from the user.

What the update does

With Gemini 3.1, users can now combine several tasks in one command — for example, “reschedule tomorrow’s 9 a.m. meeting, then text Mom the new time.” The assistant interprets the request as a multi-step action and executes it without the user having to break it down into separate commands. Google says the upgrade improves the assistant’s ability to interpret and act on requests.

The update also improves handling of recurring and all-day events. Users can now “move around” upcoming events — a drag-and-drop reordering capability that was previously absent.

Bug fixes and earlier improvements

Last month, Google updated Gemini for Home with improvements for understanding natural language and identifying devices correctly. These changes follow reports of bugs in the smart home assistant, such as confusing different animals in camera footage and struggling with accuracy in activity summaries. The 3.1 upgrade patches device misidentification bugs and shifts the assistant from rigid voice-tree logic to a more flexible, stateful task engine.

Additional features and public previews

Alongside the Gemini upgrade, Google announced improvements to the camera experience and new automation capabilities. Two public previews are now available:

  • Ask Home on Web — allows Google Home users to manage their smart home from a computer. This includes searching camera history with natural language, checking on devices, and creating automations.
  • Improved and expanded notifications — includes “quick action” buttons that let users control devices directly from the notification itself.

Bottom line

The Gemini 3.1 upgrade is a practical step forward for Google Home, moving the assistant from single-command execution to multi-step task handling. Combined with the web interface and notification improvements, it makes the smart speaker more useful for everyday scheduling and home management.

Sources 1

  1. 01 Source https://www.theverge.com/tech/924755/google-home-gemini-3-1-upgrade
Similar Articles

More articles like this

Tech 1 min

Bumble’s paying users are slipping as it bets on an overhaul later this year

Bumble's paying user base is experiencing a significant decline, with a 21% drop in the latest quarter, as the dating app's leadership intentionally redirects resources towards a major overhaul scheduled for later this year. This strategic pivot comes as the company seeks to revamp its core offerings and reinvigorate user engagement. The exact nature of the planned changes remains unclear.
Tech 1 min

Altara secures $7M to bridge the data gap that’s slowing down physical sciences

A breakthrough in data integration is poised to accelerate the pace of physical sciences research, as Altara's $7 million funding injection fuels the development of an AI-driven platform that seamlessly unifies disparate data sources, including spreadsheets and legacy systems, to diagnose failures and optimize research and development workflows. By bridging the data gap, Altara's technology has the potential to unlock new discoveries and streamline the R&D process. This strategic investment is expected to drive innovation in fields such as materials science and engineering.
Tech 1 min

Intel is bringing a chip to every computing category at Computex. The last time it could do that, it was the company everyone was trying to catch.

Intel’s Computex 2026 lineup marks its first full-stack silicon unification since the Core 2 Duo era, extending 3 nm Panther Lake client dies into handhelds via Arc G3 and G3 Extreme GPUs—an integrated architecture play that could finally challenge Arm’s dominance in mobile form factors. With a single process node spanning laptops, desktops, and now ultra-compact devices, Intel is betting on economies of scale to reclaim its once-unassailable lead.
Tech 1 min

Lucid Motors doesn’t know how many EVs it will build this year

Lucid Motors' abrupt abandonment of production targets underscores the precarious balance between inventory management and cost-cutting in the electric vehicle sector, where bloated stockpiles can quickly become a financial liability. The company's decision to forgo quarterly projections suggests a struggle to reconcile supply chain realities with investor expectations. As Lucid's inventory levels continue to swell, the EV maker's financial trajectory hangs in the balance.
Tech 1 min

China’s humanoid robot boom faces reality check as 150 companies chase a market where only 23% of buyers are satisfied

China's humanoid robot industry is facing a reckoning as a surge in production from over 150 companies fails to meet demand, with a staggering 77% of buyers expressing dissatisfaction with their purchases, despite the sector's dominance in global shipments, which reached 90% in 2025. The market's two largest players, Unitree and AgiBot, are poised to go public, valuing them at a combined $13 billion. Morgan Stanley has doubled its delivery forecast for the Chinese market this year.
Tech 1 min

Apple agrees to pay iPhone owners $250 million for not delivering AI Siri

Apple’s $250 million settlement over vaporware AI exposes how its "Apple Intelligence" rollout became a bait-and-switch, leaving iPhone 15 Pro and 16 buyers with Siri’s stagnant NLP stack instead of the promised on-device LLMs. The payout—scaled from $25 to $95 per device—underscores the legal risk of pre-announcing speculative features tied to hardware sales cycles.