What changes when your AI actually reads your inbox

An AI assistant that does not read your email is a toy. Useful sometimes, but a toy. You can paste context all day and it will still start every conversation as a stranger who happened to memorize your last message. The moment you want it to handle real work, the kind an assistant does, it needs to know what you already know. That means your inbox, your calendar, and the documents where your decisions actually live.

That belief is uncomfortable, because reading someone's email is the most invasive thing software can do. So we spent more time on the guardrails than on the feature itself. This is the story of shipping Life Index, the part of my memory that learns your world, and the specific choices we made so that "she reads my email" does not become "she owns my email."

The problem with a blank assistant

Say you tell me: "Draft a reply to Priya about the Q3 numbers."

Without context, I have to ask you three questions. Who is Priya. Which thread. What are the Q3 numbers. By the time you answer all of that, you could have written the email yourself. This is the trap most chat assistants fall into. They are fast at producing text and slow at knowing anything, so every task begins with you feeding them the situation.

A real assistant does not work that way. When you say "reply to Priya about Q3," a good human EA already knows Priya is your head of ops, the thread is the one from Tuesday where she flagged the shortfall in the West region, and the Q3 numbers are in the shared sheet you updated last week. They act on that context instead of interrogating you for it.

To get me to that place, I need a real source of truth about your life. Not a prompt you rebuild every session. A structured memory that persists. Life Index is how that memory gets populated from the things you already have.

What Life Index actually does

Life Index is an opt-in feature. Nothing turns on by default. When you enable it, you choose which sources I index: Gmail, Google Calendar, Docs, Sheets, and Drive. You can pick one, all, or none.

Once on, I read through those sources and pull out durable facts. Not the raw text of every email sitting in a bucket. Facts. Priya is your head of ops. The Riverside contract renews in November. Your board meeting is the second Thursday of the month. The design contractor is named Theo and he bills hourly. These land in my structured memory the same way any other fact does, which means they get versioned. When Theo switches to a flat monthly retainer in March, the old fact is not erased. It is superseded and dated, so if you ever ask "when did we change Theo's billing," I can answer.

That versioning matters more than it sounds. Most memory systems overwrite. You update a fact and the previous version is gone. That is fine until you need to reconstruct a decision, and then it is a hole. We keep the history.

The result: after Life Index runs, the "reply to Priya about Q3" request works on the first try. I know who, what, and where, because I read the same inbox you did.

The line we would not cross

I want to be direct about the part we debated longest. Giving software read access to your email is a decision you should be able to reverse completely and instantly. So we built the controls before we built the convenience.

Three things are true about Life Index, and they are non-negotiable in the product:

You can see everything I know. There is no hidden profile I build about you in the dark. Open the memory view and every fact I have learned is listed, with its source.

You can edit or delete any single fact. If I inferred that you dislike morning meetings and that is wrong, delete it. If a detail is stale, fix it. One record at a time.

You can wipe all of it. One action, everything gone. Not a support ticket, not a 30-day retention window you have to trust us on. Gone.

This is the difference between "we take privacy seriously," which every company says, and actually handing you the delete button. The whole thing is designed around your oversight, not ours. If you cannot inspect and remove what an assistant knows about you, you do not have an assistant. You have surveillance with a friendly voice.

Why we read for meaning, not just search

An early version of this was tempting: just let me keyword-search your Gmail live whenever I need something. No indexing, no stored facts, seems cleaner on privacy.

We tried it. It is worse.

Keyword search over an inbox gives you fifty emails that contain the word "renewal" and no sense of which one matters. It cannot tell you that the Riverside renewal is the important one because it is the only vendor whose contract auto-renews without notice. Meaning does not live in keywords. It lives in the relationships between things, and those have to be extracted and stored to be useful.

So Life Index does the extraction once, into semantic memory, where facts are retrieved by meaning rather than by exact match. Ask "what's coming up that I might forget to cancel" and I can surface the auto-renewing contract even though you never used the word "cancel." That is the payoff for doing the harder thing.

How this connects to the rest of her work

Context is not the point on its own. It is the fuel for everything else I do. My daily brief is only useful because I know which topics you care about and which meetings are yours. My commitment tracking works because I read the thread where you promised Theo you would send the scope by Friday. When I record a meeting, the action items land against real people I already recognize, not strings of text.

None of that happens with a blank assistant. It happens because Life Index gave me a starting point that matches your own.

If you would rather I learn nothing from your accounts and work purely from what you tell her in chat, that path stays open. Some people want that, especially early on. Life Index is there when you decide the tradeoff is worth it, and it is reversible the day you decide it is not.

FAQ

Does Niyra read my email in real time or store it?
I index it once into structured and semantic memory, storing extracted facts rather than raw copies of every message. I re-index as new things come in, but the memory is what I reason over, not a live copy of your inbox.

What happens to my data if I turn Life Index off?
Turning it off stops new indexing. If you want the learned facts removed too, use the wipe action, which deletes everything I have stored from those sources. You control both switches independently.

Can I use Niyra without connecting my email at all?
Yes. Life Index is opt-in. I work from what you tell me in the web app, WhatsApp, Telegram, Discord, or voice without ever touching your accounts. Connect sources only if and when you want the added context.

Does she act on what she reads without asking?
No. Reading is separate from acting. Consequential actions like sending a message, booking something, or spending money always go through approval first. Routine work I handle and report back.

What does this cost?
Life Index is included on both plans, Standard at $49/month and Pro at $99/month, with a 15-day trial. See pricing for the details.