Google Gemini Spark AI assistant launches with 24/7 cloud agent

Gemini Spark, Google’s new always‑on AI assistant, became generally available after its debut at the May I/O conference, offering to run tasks like email summarisation, shopping‑deal research and weekend‑planning from the cloud without keeping a laptop awake [1].

Agentic AI for everyday work

Google positions Spark as a “agentic” assistant that can act on behalf of users across Gmail, Calendar, Docs, Sheets and Slides, handling routine digital chores without the user needing to keep a machine running [1]. In practice, early‑access testing shows Spark can pull together shopping lists with coupon suggestions, generate packing lists based on weather data, and summarise newsletters from a user’s inbox. The assistant, however, still lacks integration with Google Keep, forcing it to fall back to Docs or email for note‑taking [1].

How Spark stacks up against rivals

Spark’s cloud‑only design contrasts with competitor OpenClaw‑style agents that require a continuously running local instance, a point Sundar Pichai highlighted at I/O as a “you can close your laptop” advantage [1]. While the feature set mirrors existing productivity tools, the real differentiator is the ability to run tasks continuously in Google’s data centres, reducing local resource consumption. The assistant’s conversational abilities are similar to those seen in Google’s Gemini‑powered smart speaker, which introduced “Continued Conversation” to let users issue follow‑up commands without a wake word [2]. That speaker, priced at $99.99, demonstrates Google’s broader push to embed Gemini across hardware and software, though the speaker’s audio hardware (single 58 mm driver) received mixed reviews [2].

Early user experience and limits

Testers reported that Spark can locate local deals and suggest coupon stacking, though some promo codes proved invalid, highlighting the need for human verification [1]. The assistant also struggled with incomplete requests—failing to fetch price‑drop alerts at a user‑defined frequency and omitting cost details for summer‑activity searches [1]. These gaps suggest that while Spark is functional for structured, Google‑centric tasks, it still requires refinement for broader consumer use.

What to watch

• Full rollout timeline – Google has not disclosed when Spark will be enabled for all Workspace users beyond early access.
• Integration updates – Adding native support for Google Keep or other note‑taking apps could close a key usability gap.
• Competitive response – Rival AI agents may accelerate local‑instance offerings to counter Spark’s cloud‑only model.

Gemini Spark marks Google’s first major consumer‑focused AI assistant built on its Gemini model, extending the company’s push to make AI conversational and always‑on. Whether the convenience of cloud‑hosted tasks outweighs the current functional gaps will become clearer as Google expands integration and competitors react.