Google is testing a new version of CC, a Google Labs experiment that lets a household share an AI agent. Up to six people can feed it their data so it can make plans and finish tasks for the group. That is a family-sized product, not a workplace one.
CC started in 2025 as a personal tool. The original project eventually became Gemini's Daily Brief, which reads a Google account and turns it into action items and suggestions. The new CC is aiming at the same job, only for a household instead of one inbox.
From a 2025 personal brief to a six-person agent
Ars Technica first reported the family pitch and the six-user cap. Multiple family members share data to help the agent make plans and complete tasks. Six is a Family Link-sized household, not a team at work. Someone else's calendar, someone else's lists, one agent trying to make dinner or a weekend trip add up.
Google's models may not lead by most measures right now. The company's real advantage is your data. If you are deep in Gmail, Calendar, Maps, and the rest, Gemini already has a lot of context on you. A family agent is a way to spread that context across more than one person. It only works if people are willing to pool the data Google already holds on them.
Labs is where the company tries this kind of thing before it becomes infrastructure. CC is still an experiment, not a default Gemini feature. Treat it that way. Features that start in Labs sometimes graduate. They also sometimes vanish after a blog post and a waitlist.
Personal assistants have been promising to just handle it since the first-generation Google Assistant and Siri years. Daily Brief was a more honest product: it reads what you already store and nags you with a list. Family CC is Daily Brief with more passports in the account. The hard part is not summarization. It is whose flight, whose dietary restriction, whose after-school pickup wins when the agent has to pick.
The advantage is Gmail, not the leaderboard
A family agent that cannot see Calendar is a chatbot with extra steps. One that can see everyone's Calendar is a household operating system with a privacy edge. Google is pitching convenience. Parents will want to know what the agent can see, who can add a member, and whether a teenager's mail is in the pile.
Apple's household sharing and Amazon's Alexa household profiles are the obvious comparisons, and they are not the same product. Those systems share subscriptions, smart speakers, and a few permissions. They do not, as a default, let one model rummage six inboxes to plan a Saturday. Google can attempt that because the accounts already live on its servers. That is the advantage. It is also the reason to pause.
I would not turn this on for a household with a kid who has a Gmail they treat as private. I would consider it for two adults who already share a calendar and a grocery list, if Labs gives a clear log of what was read. Without that log, you are trusting a summary of six lives to a model Google does not even claim is in the lead.
DeepMind is busy standing up an institute to air disagreements about AGI at the research end of the same company. Family CC is the consumer end: less philosophy, more access to the Tuesday of your actual household. Both are Google trying to turn we have the context into a product. Only one of them will see your kid's calendar.
What to check before you pool the accounts
Ask who can invite the sixth user, and who can kick them out. Ask whether Daily Brief and family CC share memory, or whether a personal brief leaks into the group agent. Ask what happens when two people tell it to book different dinners. If Labs cannot answer those in the experiment UI, wait.
Practical availability is Labs-only for now. There is no public claim in the source material that this is rolling out to all Gemini users, and inventing a date would be a mistake. If you are not already in the Google account stack, this product is not for you. If you are, the six-person cap is the whole design: nuclear family, not a group chat with cousins.
Robotaxis and living-room gadgets are not family agents, but they are the same test of whether Google's hardware-and-data story survives contact with other people. Waymo's Singapore timeline is a long supervised ramp before paying riders show up. Valve's Steam Frame is engineers arguing about whether a device should stand alone or lean on a pipe. Family CC leans entirely on the pipe. The pipe is your household's mail.
Sharing inside a family is a product idea people already accept for photo albums and location. They have not accepted it for an agent that completes tasks. Completing tasks is the line. Planning a trip is one thing. Sending the email is another. Until Google says which side of that line CC sits on, keep it on the plan-only side yourself, and do not feed it every account in the house just because the cap is six.