Generative AI has driven the cost of “high-quality harassment” toward zero. alink hides your email and WeChat behind a rule-aware agent gateway — noise is quietly stopped, worthy requests are politely let through, and the final call is always yours.
Put al.ink/name in your email signature, homepage, or meeting card. Outside reach no longer lands in your inbox — it submits a structured request per your contact contract, to be verified, filtered and triaged before it can reach you. The link is never welded on; revoke it anytime.
Pick a policy template (investor / founder / office hours / private) and tune the admission contract: which requests are allowed, what context is required, how fast you respond. What you preview is what ships; publishing gives you a permanent link.
No sign-up, no agent needed on their side: they fill a structured form per your contract, and the AI asks at most two follow-ups on vague input. Before submitting, they are clearly told “your request will be handled first by an AI agent.”
Requests land in your Assistant Inbox by tier, with a summary and the reason for the tier. Approve to deliver the contact channel per contract; decline to send a gracious reply in your preset tone. Approved requests settle into relationship records automatically.
Approvals and authorizations are always confirmed by you in person; the AI has no right to overstep. This is not a setting — it is a boundary written into the protocol. The gap in the ring is its shape.
There is no central “who knows whom” table in the system. Relationship slices are encrypted and sharded per user; any cross-user query must be authorized by both sides, computed transiently, and left in an audit trail.
Authorizations, credentials and relationships can all be downgraded, cooled down or revoked — in both directions. Every agent action is written to a verifiable audit log; data exports in full, and deletion is a cryptographic erase.
alink never impersonates a human: the AI takes part in handling and replying, and the other side always knows. Those declined get a gracious reply — not humiliated, and unable to probe your rules.
What you pay for is protection and identity, not reach privilege. “Pay to raise pass odds, jump the queue, buy exposure” is vetoed by a red line — the sender is free forever, no sign-up.
This is a security product: a lapsed subscription only downgrades to rule-layer filtering + a daily digest — it never lets requests through over unpaid dues. Your xid is a land deed, free forever; a handle is only a house number, redeemable anytime within the grace period.
The admission contract is the soul of the gateway: you declare “how I am willing to be contacted” in a machine-executable form — which requests are allowed, what context is required, how fast you respond, what is delivered on pass. No abstract concepts to configure; pick a template and publish. Every verdict traces back to a specific clause in the contract.
The contract is versioned: changing rules produces a new version, and historic requests are always judged by the version at submission time — explainable to you, auditable for the record.
What you buy is protection and an identity asset, not reach privilege — the sender is free forever, no sign-up. Your xid is free forever and non-recyclable; a lapsed subscription only downgrades, never drops the guard.
The ring is the link, the gap is authorization, the dot is the person guarded at its center. The dot in the domain and the dot at the ring’s heart are the same dot: your final decision.
The first batch opens to those hit hardest: investors, founders and open-source maintainers who get 50+ cold reaches a week.