Yes, for some errands, as of July 30, 2026. Gemini Spark can now use your logged-in Chrome accounts and saved passwords to handle web tasks like scheduling apartment viewings or researching and starting a flight booking. Google says it guards against prompt injection and hands payment steps back to you to confirm.
Can Gemini Spark actually run your personal errands now?
That's a specific, dated update, not a description of the whole Gemini app. Worth remembering if you've been burned by “AI assistant” headlines before: this piece sticks to one narrow, dated change, not a sweeping claim. Everything below is based on Google's own documentation and independent reporting, not a hands-on test on our end.
What is Gemini Spark, and what changed on July 30, 2026?
Gemini Spark is a personal AI agent inside the Gemini app. Google describes it as a “24/7 personal AI agent designed to proactively manage tasks and help you navigate your digital life, all under your direction” (blog.google, May 19, 2026). It launched to trusted testers that week, then opened to U.S. Google AI Ultra subscribers as a beta.
The July 30, 2026 change is narrower and more useful: Spark can now use your logged-in Chrome accounts and saved passwords to handle “tedious web errands,” according to Google's own post that day. Before this, Spark worked through a remote browser, separate from your actual logged-in sessions. Now, with your permission, it can act inside Chrome on your desktop, using accounts you're already signed into.
Google organizes Spark's work around three building blocks: a task (the high-level goal you give it), a schedule (hands off a recurring task so it repeats automatically), and a skill (a set of reusable instructions). You will meet all three below.
Worth saying plainly: Google isn't inventing this category. OpenAI's Operator has attempted similar web-browsing errands for a while. Spark's July 30 update is Google catching up, and in some ways extending, what agentic assistants already try to do.
What personal errands can Gemini Spark actually do right now?
So what does that “catching up” actually look like day to day? Google names two examples directly: “scheduling viewings for apartments you've saved” and “researching flight options and starting the booking process” (blog.google, July 30, 2026). It's the same job as the plain way to plan a trip with AI, just running one layer deeper, inside your own browser.
Beyond those two, Spark connects to Gmail, Google Calendar, Docs, Sheets, Slides, Drive, Google Keep, and Google Tasks, per Google's support documentation. For personal life, that means things like drafting an email reply, pulling together a packing list in Docs, or setting a reminder tied to your calendar.
Independent commentary backs this up, which matters if you're not the type to take a company's own blog post at face value. One X post from @WesRoth, a day after launch, put it simply: “the agent can use logged-in accounts and saved passwords to handle multi-step web errands” (July 31, 2026).
Another, from @ai_for_success, listed the kind of jobs people were already trying: “tracking travel bookings, finding events, auditing subscriptions, organising emails, or drafting replies for you” (August 1, 2026). Neither is Google's own claim. Both are independent confirmation of the update's real scope, not just marketing copy.
One thing this article won't cover: Spark also connects to business tools like Salesforce, HubSpot, Asana, and Linear through MCP. That's real, but it's a work-tool story. This guide sticks to your own life, not your job.
The privacy trade-off: what “using your saved passwords” really means
Let's be direct about this, because it matters, and because “trust me” isn't good enough when passwords are involved. Giving Spark this permission means it can sign into sites the way you already do, using credentials Chrome already has stored, and click through multi-step processes on your behalf.
That's a real shift, not a cosmetic one. Up to now, an AI assistant answered your questions.
Now it can act inside your accounts: airline, apartment-listing site, maybe your email. You're extending trust, not just convenience, and it's fair to sit with that for a second before you flip the switch.
Android Authority's Shimul Sood put the trade-off plainly: “At the end of the day, it all comes down to how much convenience you're willing to trade for privacy” (July 31, 2026). Sood also admitted he'd personally hesitate: “I'm still not convinced I want an assistant logging into my personal accounts on my behalf.” That's one reporter's opinion, not a universal verdict, but a fair gut check before you turn this on.
Google built in guardrails for exactly this reason. Here's what they are.
The guardrails Google built in (and why your supervision still matters)
Google states Spark works “while protecting against threats like prompt injection and keeping you in the loop on sensitive actions, such as payments, by handing the task back to you” (blog.google, July 30, 2026).
In plain terms: prompt injection is a hidden instruction planted on a webpage, designed to trick an AI agent into doing something you never asked for, like sending your data somewhere it shouldn't go. Google says it's building defenses against this specifically.
The second guardrail is simpler: anything involving actual payment gets handed back to you. Spark can research and set things up, but it stops short of clicking “confirm purchase” without you.
Google's support page adds more: you can pause or take over a task mid-run, watch progress in a work panel, delete remote browser and computer data, or just turn Spark off entirely.
Guardrails lower the risk. They don't remove it. That's exactly why the decision table below matters.
How to set up Gemini Spark, step by step
No mystery here. Per Google's own documentation, here's the process:
- Confirm you qualify: 18 or older, a personal Google Account (not work or school), Keep Activity turned on, and a Google AI Pro or Ultra subscription.
- Open the Gemini app or gemini.google.com.
- Go to Menu, then Spark.
- Describe your task in plain language. Try one of Google's own examples as a template: “find apartment viewings for the places I've saved” or “research flight options for [your trip].”
- Submit it, then check the work panel. Spark will flag any step where it needs your input, especially anything involving payment.
- Know where the controls live: you can pause a task, take it over yourself, or delete remote browser data if you change your mind.
Spark is not a free feature. The cheaper of the two tiers that include it, Google AI Pro, runs $19.99 a month. Google AI Ultra runs $99.99.
On price: Google AI Pro runs $19.99 a month and includes Spark plus Chrome auto-browse (US only), per Google's own pricing page (one.google.com, checked August 2026). Google AI Ultra runs $99.99 a month. Neither is free, so don't expect this to sneak in as some hidden freebie.
On availability, keep two rollouts straight. Chrome errands are U.S.-only for now, with other regions planned later. General Spark access expanded to over 160 more countries on July 30, 2026. But per Google's own support page, Spark itself still excludes the European Economic Area, Nigeria, Switzerland, and the United Kingdom, regardless of tier.
Which errands to hand over now, and which to still do yourself
Here's the plain breakdown, task by task.
| Personal errand | Hand it to Spark now? | Why |
|---|---|---|
| Apartment-viewing scheduling | Yes | One of Google's own named examples; low-risk, no payment involved |
| Flight research and starting a booking | Yes, for research | Spark researches and starts the booking; you confirm the purchase, much like you'd keep an eye on flight prices without doing it yourself |
| Subscription auditing | Worth trying | Matches documented Gmail/Calendar access; still check the results, similar to already using AI to cancel subscriptions you don't need |
| Email triage and drafting | Worth trying | Gmail connection is built in; review drafts before sending |
| Anything ending in a final payment | Do it yourself | Google hands payment steps back to you by design, for now |
The line to remember: treat Spark as the research-and-setup layer, not the checkout layer. That still covers the everyday admin that eats 20 to 30 minutes at a time, not a full handoff of your digital life. If that's the kind of time you're trying to claw back after work, this is worth the ten minutes it takes to set up.
Frequently asked questions
Is Gemini Spark available?
Yes. Spark launched to Google AI Ultra subscribers in May 2026 and now includes Google AI Pro subscribers too. Chrome auto-browse errands are U.S.-only for now.
Is Gemini Spark free, or do I need to pay for it?
No. Spark requires Google AI Pro ($19.99/mo) or Ultra ($99.99/mo), per Google's own pricing page. There's no free tier.
How do I get started with Gemini Spark?
Open the Gemini app, go to Menu, then Spark, and describe a task in plain language. You'll need a personal Google Account, be 18 or over, and have Pro or Ultra active. See the setup steps above.
Can you use Gemini as an AI agent?
Yes. Spark is Google's agent layer inside the Gemini app, running tasks, schedules, and skills in the background, including, since July 30, 2026, errands using your logged-in Chrome accounts.
How to use Gemini AI as personal assistant?
For everyday chat and drafting, just talk to Gemini directly. For multi-step errands like apartment viewings or flight research, use Spark and follow the setup steps above.
Which AI agents are best for personal use?
It depends on the job. Spark, ChatGPT, Claude, and OpenAI Operator each handle personal tasks differently: Spark and Operator lean toward taking actions, ChatGPT and Claude toward drafting and research. For a deeper comparison, see the best AI assistants for personal life.
If you want plain updates like this one when a new AI feature actually changes what you can do day to day, the newsletter covers it without the hype.
The honest bottom line: who should try Gemini Spark now, and who should skip it
Try it if: you already pay for Google AI Pro or Ultra, you use Chrome as your main browser, and you have a real repeat personal errand, like ongoing apartment hunting or regular travel research. Start with one low-stakes task, not your whole digital life. Treat it as a trial run, not a leap of faith.
Skip it for now if:you don't already have Google AI Pro or Ultra (don't subscribe just for this), you're outside the U.S. waiting on the Chrome-feature rollout, you're not comfortable with an AI agent holding your logged-in sessions and saved passwords, or you only have occasional errands where setup isn't worth the trouble.
This feature was 12 days old at the time of this writing, still wet paint by tech standards, and unproven at scale. Try it on one small thing before trusting it with more.
For the broader picture beyond this one feature, see the full range of personal AI tasks. If the password trade-off isn't for you, a lighter-touch option like Siri covers a lot of the same everyday ground without it.