How Well Do You Want Google, Apple and Microsoft to Know You?
Google launched Personal Intelligence in January 14, 2026, for US Google AI Pro and Ultra subscribers, and it evolved out of an earlier feature called Personal Context, which learned from your past Gemini chats. You opt in and connect Google apps, currently Gmail, Photos, YouTube, and Search, and Gemini pulls from them to personalize answers. It's since expanded to AI Mode in Search and Gemini in Chrome, with a free-tier rollout starting in March. I have Google service. I do not recall opting in for this level of scrutiny. The opt-in likely came with some kind of change in Terms of Service that we all routinely click a check box, agree, and move on.
Microsoft 365 Personal, Family, and Premium subscribers get Copilot in Word, Excel, PowerPoint, Outlook, and OneNote: drafting in Word, analysis in Excel, thread summaries in Outlook, plus Copilot in Teams and Designer, with higher usage limits on the new Premium tier. The Copilot app remembers preferences and details from your chats, and the "Microsoft usage data" setting extends personalization to your Bing, Edge, and MSN activity. It’s called recall on Copilot+PC. Recall takes snapshots of your screen every few seconds, indexes them locally on the NPU, and lets you search your own history in natural language: "the PDF about healthcare trends I opened Monday." Copilot can query your Recall timeline directly, so it can answer from what you were actually doing across apps, documents, websites, and chats . Microsoft says that snapshots stay on the device, encrypted, and are never sent to its servers.
Think about this, recall is what context reconstruction looks like. Microsoft can't persist meaning across Word and Outlook and your browser, so it persists pixels instead. Thousands of screenshots a day. Your bank balance when it was on screen. The message from your attorney. The medical portal you had open. All of it photographed, in case you ask about it later. And it still wouldn’t know that an email sent to your friend John is the John in the spreadsheet you have been working on with him. It knows what your screen looked like at 2:14 on a Tuesday and every second before and after.
Apple confirmed in January 2026 that the next generation of Apple Foundation Models is based on Gemini, reportedly paying around $1 billion a year for a custom build running inside Private Cloud Compute . On-device tasks like quick personal-context lookups and on-screen awareness run on Apple's own models on Apple Silicon; heavier reasoning routes to the Gemini-powered cloud . So the company selling "personal context on your device" is renting the intelligence from the company that sells ads against your context. Apple kept the interface and the privacy plumbing; the mind is a vendor's.
The context layer is App Intents, and it's clever. Apple did not build a big model over your data. It built a contract for developers. Apps describe their content as "entities" against system-defined App Schemas, donate those entities to a Spotlight semantic index, and expose actions as intents . That gives Siri three things: it can find your stuff by meaning rather than keyword, it can act across apps ("email the people in this event"), and onscreen awareness lets it resolve "this" and "that" from what's on your screen .
Let's take a moment to think this through:
Add up what you have just handed over. Google now reads your correspondence, your photographs, your search history, and everything you have ever asked for, and it calls the result Personal Intelligence. Your search history is the richest context you can produce. You type things into that box at midnight that you would not say out loud to your spouse. Microsoft is photographing your screen every few seconds and storing the film on your own machine, which means the record of your financial life, your legal matters, and your health now exists as a single searchable archive, and the only thing standing between that archive and the world is one company's engineering and another company's promise. Apple, the one that built a brand on privacy, decided its assistant was not smart enough and wired your questions into Google's models to fix it.
Notice that none of this is a breach. No laws were broken. This is the product working as designed, with your consent, which you gave in a checkbox. Each company will tell you the data is encrypted, on-device, opt-in, reviewed. All of that can be true, and it does not touch the actual issue. The actual issue is the incentive underneath the feature. Google's business is advertising, and an advertiser's dream has always been to know what you want before you do. Microsoft sells to enterprises, so the deep version of its context layer goes to your employer, not to you. Apple sells devices, so your context is welded to the hardware and dies with it. Every one of them will know you extraordinarily well. None of them work for you.
That is the question the title asks, and it is not rhetorical. How well do you want these companies to know you? Because the answer they have chosen on your behalf is: completely, permanently, and on their terms. Google will know what you ask. Microsoft will know what you do. Apple will know what you mean, and it will rent Google's brain to figure it out. The most intimate record ever assembled of a human life is being built right now, one terms-of-service update at a time, and the people building it are accountable to advertisers, enterprise contracts, and hardware margins. In that order.
A context layer is going to exist. That fight is over, and these three announcements settled it. The only question left is whose it is. A layer that knows this much about you should have exactly one customer, and wouldn’t you like the customer to be YOU.
If that is not enough, the implicit promise of all these platforms was convenience, ease of use and they were free. The frontier of AI and the integration into your personal lives and families is opaque, for now it all seems helpful, fun and for the most part, harmless.
But is that all true?
Free is never free:
Search sold your intent. Social sold your attention. AI context sells your meaning.
Each generation of free went deeper into your life. Each bill took years to arrive. This one goes deepest.
Convenient is the mechanism for dependency
These features seem genuinely useful on day one. That is how dependency gets built.
Nobody hands over their life to a system that annoys them. You hand it over one small kindness at a time.
Harmless compared to what and measured when?
It took fifteen years to learn what free social media did to teenagers. The answer arrived as a surgeon general's advisory.
These systems are fifteen months old and already see more than Facebook ever did. Facebook watched what you posted. These see what you meant.
Opacity is the strategy and not a phase.
The collection is priced later. Google's ad business, Microsoft's enterprise contracts, and Apple's hardware margins will each decide what your context is worth to them
You found out what free search cost you after the fact. This time what you paid with cannot be deleted. It is the resolved meaning of your life.
The promise was convenience. The terms were always going to be set by whoever owned the layer. It was never going to be you.
And, will these so-called services ever be fully free from an actual dollar cost point of view, I think not and this is why:
The computation is new and it is real. Before AI, your email cost Google almost nothing to store and serve. Now every message, every photo, every search gets reasoned over by a model. That cost did not exist five years ago, and it does not disappear because the price per inference falls.
Continuous understanding costs more than occasional answers. A chatbot you ask twice a day is cheap. A system that understands your life all day, every day, is not. Usage is moving from the first to the second, and the aggregate bill grows even as unit costs fall.
"Free" tiers are subsidized, and subsidies have a payer. Google pays with advertising, which means advertisers are the customer. Microsoft folds it into licensing you already pay for. Apple builds it into a $1,200 phone. You are paying. The invoice just arrives addressed to something else.
The meter is already running. Google put Personal Intelligence behind AI Pro and Ultra subscriptions first. Microsoft raised the price of Microsoft 365 and called the increase Copilot. Apple is reportedly planning paid Siri tiers. Free was the introductory offer, not the business model.
The features worth having will be the features that cost. Basic summaries stay free. Deep context, the kind that knows your whole life and acts on it, is expensive to run and will be priced accordingly. The free version will be the demo.
And the unpaid balance is you. Whatever they cannot cover with subscription revenue gets covered the old way, with your attention, your behavior, and your data. There is no version of this where a multi-billion-dollar context infrastructure runs on your life's details out of generosity.
Therefore ask yourself; will these services remain free from any kind of paid tier for more functions? History shows no and it is already happening with loss leader AI products from all the major companies; Open AI, Antropic and Google. Their free tiers go to tier one: ~$20-35/month, tier 2: ~$100-20/month and tier 3: ~$200-280/month.
Is there a better way? There is. Not only are there cheaper ways to deliver context and meaning across applications and make a person's life easier, the whole thing can be privately owned by the consumer, the small business owner, or the enterprise. No company would own any entity's data or, more importantly, the meaning and context of that data. I cannot overstate how important it is to never give that up. What social media did with your data and its monetization will pale next to what this information is worth for profit and power, across every domain of our lives, from the mundane day to day to global politics. Never before in human history could data be assembled with the richness this new world of AI allows. Whoever holds that assembly holds you and your agency.
That better way is the PSL/i. PSL/i is an independently owned digital storage layer for any entity's data and the meaning associated with all that data. The paradigm for AI storage is no longer folders and files. It's data in tables, knowledge graphs, vector indexes, memory cache and more. All together these new AI tools construct a persistent, deterministic digital storage layer of your complete life as represented by all your data and the collation of that data across all your apps. That collation, assembled by AI and held in context, is what creates the richness: meaning, commitments, decisions, and your intent. It's your agency in digital form.
If that weren't enough to get your attention, there is so much more. The PSL/i has been architected and scoped to be built now. The PSL/i is a product protocol. What I just explained was how your data gets stored and held and why it's yours. The PSL/i product ships with a Navigator (not a chatbot). The Navigator is used by a user to state how they want to use their data and for what purpose. That purpose can be anything a computer user does today. Applications aren't necessary because applications are just "tool" calls that operate inside the PSL/i product. A user can write emails, create spreadsheets, do personal accounting, book hotel rooms, order from Amazon, anything that is done now can be done through the PSL/i. As mentioned, apps aren't necessary, browsers aren't necessary, and neither is search. They are "tool" calls that the PSL/i presents to the user if those tools satisfy the intent of what the user expresses through the Navigator. The PSL/i can then reason over all that structured digital storage layer to do multi-module work across different domains. You may ask, how does the PSL/i handle such complex tasks? The PSL/i is a purpose-built modular software product built on protocol. Different domains such as communications, finances, time, accounting, legal, shopping, really any domain would be handled by purpose-built Lenses (instead of apps). A Lens has its own schema, its own ontologies, its own rules, even an attached small language model (SLM) if necessary to connect your own data, tools, or the internet to accomplish tasks and work within that specific Lensed domain. This is how AI works for the consumer, the small business owner or the large enterprise and not for a huge conglomerate looking to sell or advertise against your information.
In closing, there is a choice. The PSL/i is scoped and ready to be built. The investment is small against capex of $3T and counting in the AI space. How small? 0.00033% of $3 trillion.
Before the mainframe there was ENIAC, after the mainframe came the PC. ENIAC used punch cards, the mainframe used magnetic reels, the PC started with floppy drives, and moved to on-device hard drives and solid-state drives. Then came local area networks and office connected PC systems, and from there AOL, and dial-up internet. Applications weren't apps, they were programs run locally and privately. The internet became useful with the world wide web protocol. Yahoo pioneered search and Google made it useful and blossomed. As internet connectivity speeds increased so did its usefulness. SaaS and cloud based computing supplanted on-device programs and enabled cross-device usefulness. The planet is now strung with millions of miles of fiber optic cable with data traveling at the speed of light. We are now and forever physically connected.
What has been consistent across technology innovation is that no incumbent remains the incumbent forever, especially if there is a superior protocol and product on that protocol. Their instinct is to protect their walled empires, and that is happening now. Protecting themselves is not protecting you and your businesses. It is not about you; it is about them. It would be foolish not to recognize it for what it is. Never in human history has a technology like AI presented such an opportunity to improve people's lives across every aspect, from their families to their professions and the opportunities within those professions. No company, incumbent or otherwise, needs to suffer, no stock price needs to fall, no business needs to fail. The only thing that needs to happen is to make the right choices that benefit everyone and to ensure that every entity's data and meaning remain theirs; their intellectual property with their value . That time is upon us. Choose wisely.
