A copilot that answers to the person in the chair.

Most AI products are built to remove the human — automate the task, close the loop, send you the result. LivePair is built the other way: it watches your screen, hears the room, and hands you the answer while the work is still open — the human stays in the chair, and the copilot makes that human sharper.

A live session pairs you with Joy Alive — it sees your screen, hears the room, and holds the thread of what you're doing — so you can stay in the work while it handles retrieval, recall, and the second pair of eyes. The point isn't a smarter model. It's a more capable you.

LivePair's runtime is organized like a synapse, not a chatbot: continuous perception feeding a persistent internal state — beliefs, goals, uncertainty, a model of itself — which decides what to surface and when. It was designed for presence, not prompts.

• Persistent state — the session has a memory and a position on what it's seen, so it never starts cold mid-conversation.
• Exposed uncertainty — the agent reports confidence, not just answers. A guess is labeled a guess.
• Proactive surfacing — it speaks when something matters, not only when spoken to.

That structure exists for one reason: real work doesn't pause to phrase questions. A copilot that only answers questions arrives too late to be useful.

• No covert steering — the agent never nudges you toward an outcome you didn't ask for. Its suggestions are visible and labeled.
• No fabricated confidence — uncertainty is rendered in the HUD, not smoothed over.
• Your data isn't the product — sessions aren't sold, mined for ads, or used to train models for other customers.
• No dark patterns — no expiring credits, no subscriptions you forget, no engagement traps. You pay for what you use.

LivePair is stewarded by its operating company — there is no governance-token theater and no voting on safety rules. Decisions that affect users are made in the open and documented here:

• Published operating limits — capacity caps, spend ceilings, and rate limits are public policy, not silent throttles.
• Model disclosure — the base model powering live sessions is named publicly and updated in the open as it changes.
• Plain-language terms — privacy and terms are written to be read, not buried.

LivePair runs on pay-as-you-go credits and time-based sessions — priced at a disclosed markup over provider cost, never on engagement or attention. No subscription traps, no expiry.

Every account also accrues points from day one — a public accounting of who built this with us. If LivePair launches a network token — a design target for later, not today — the model is a fair launch:

• Fixed genesis supply — 100,000,000 — no presale, no insider allocation, no venture discount. Distribution is earned by usage, not bought.
• Points → allocation — the people who used the product hold the claimable weight. The points ledger is the snapshot.
• Utility, not speculation — stakers share a daily compute-credit pool funded by a fixed share of platform margin. Staking yield is backed by real margin, so the token's demand is the product's demand.
• Programmatic revenue burn — a fixed share of platform margin is routed to a burn reserve that buys and burns supply on a schedule, verifiable on-chain. Margin grows → burn grows → the claim tightens.

This is a design target, not a promise — no token exists today, and nothing here is an offer or solicitation. Points are an in-app score with no cash value.