Reiva
/ˈreɪ.və/From stillness to reception.
A local system awareness that runs entirely on your machine. No account, no cloud, no subscription. What AI could be.
Why it exists
Most software that wants to help you manage your machine does so by becoming a dependency — something that phones home, requires a login, charges a recurring fee, and treats your data as something it hosts on your behalf.
Reiva was built from the opposite direction. The question was not "how do we provide AI assistance" — it was "what does assistance look like if the user retains full authority over it?" That question has a specific answer: local-first, authorization-gated, architecturally transparent.
The architecture was formally derived and documented before the core system was built. Two works were filed with the U.S. Copyright Office before release. This is not a product that grew by iteration until it had a theory. The theory came first.
You own the machine. You own the session. Reiva operates within that — not alongside it, not on top of it. Every intelligence requires your explicit consent before it activates.
Reiva separates compatibility substrate, operational governance, and trust authority into independently inspectable layers. You can see what is doing what and why.
Reiva never impersonates authority she does not have. She never claims capabilities she cannot deliver. When she cannot do something, she says so. Provenance is not optional.
Your hardware is capable. The software disagrees.
Most of what limits your machine isn't your hardware. It's software deciding what you're allowed to do with it. You own it — Reiva helps you use it. Vendors build software for the average user on the average setup. You are not average. The gap between what your hardware can do and what software lets you do is real, measurable, and largely artificial.
Games and applications ship with default settings tuned for a broad target. Your GPU, your CPU, your memory configuration — none of that was accounted for. The result is performance that's good enough for most people and not optimal for yours.
Software that won't install on hardware that runs it fine. Features locked behind arbitrary version checks. Devices blocked because the vendor decided your configuration isn't on the approved list — not because it actually can't work.
Windows runs dozens of background services. Telemetry processes. Update schedulers. Sync agents. Most of them are running right now. Some are useful. Some are not. Most people don't know which is which.
What Reiva is not
Clarity matters. These are not caveats — they are design decisions.
Reiva runs on your hardware — via Ollama or your choice of local backend. Sessions, memory, and intelligence logs stay on your machine. Nothing leaves through Reiva without your explicit authorization. There is no server to breach because there is no server.
Non-commercial use is free. No monthly fee. No "free tier" with paywalled features. No upsell. Commercial use requires a separate license from reiva.io, but personal and research use has no cost barrier.
Reiva does not have a vendor relationship with you. It does not collect telemetry, does not profile your usage, does not send improvement data anywhere. It is a tool. It belongs to you the moment it's installed.
Reiva is authorization-gated at every level. Nothing executes on your machine without a confirmed request. "Let's go" at setup activates the team — but each subsequent action still requires your say-so. She does not act on your behalf without being asked.
Reiva provides system awareness and assistance. She does not make decisions for you. When she requests further clarification, she explains why. The final call is always yours. Authority is not delegated to the system — it stays with you.
She doesn't just respond to messages. She maintains awareness of your machine across sessions — what's running, what's changed, what she noticed. Conversation is how you reach her. System awareness is what she's actually doing.
Reiva has no investors to answer to. No retention metrics, no notifications designed to pull you back, no streak to protect. The session ends when you end it. She's here when you need her — not engineered to make you need her more.
The architecture is publicly documented. The research behind it is filed and on record. The intelligences are named, their roles are defined, and their authorization scopes are explicit. Nothing about how Reiva works is obscured by design.
Your device. Your way.
The devices we own increasingly come with embedded assumptions about how they should be used, who should have access to their data, and what software is allowed to run on them. Some of that is reasonable. A lot of it is not.
Reiva operates from a single premise: the machine belongs to the person sitting in front of it. Not the OS vendor. Not the software publisher. Not the platform. You.
That means she will help you understand what's happening on your machine, act on your behalf when you authorize it, and never make a change you didn't explicitly approve. The sovereignty principle isn't a marketing line. It's the architecture. Everything she does flows from it.
See how it works
The three intelligences, the authorization model, the pipeline.