How Loxley AI autonomous mode works: strategy waves, risk levels, how it interacts with auto-buy (shadowing), known-actor blocking, kill switches, and why the strategy lane never double-fires.
Beyond rules-based auto-buy, Loxley AI supports an AI autonomous mode: a strategy layer that evaluates launches and initiates purchases itself. This document describes the decision model and — more importantly — the guardrails that wrap it.
Instead of reacting to every alert, the AI lane operates in deliberate evaluation waves: candidate selection from live launch flow, scoring against the full safety stack (contract, locks, snipe concentration, hooks, simulation), position sizing, entry, and managed exits through the standard sell engine. The strategy can decide; it cannot bypass the safety gate.
Strategy behavior is bounded by an explicit risk level (conservative to aggressive) that tunes maximum position size, minimum liquidity, and scanner thresholds. A conservative profile treats more of the launchscape as untouchable; an aggressive one permits smaller, newer positions — within the same hard veto rules. Risk level changes apply to future waves, never retroactively to open positions.
If a user runs AI autonomous mode and classic auto-buy on the same channel, double execution would result. The engine resolves this with lane-level shadowing: when the AI lane is active for a lane, plain auto-buy is fully silent there — no buys, no duplicate member messages, no second scans billed. The strategy wave becomes the single purchasing voice for that lane; other lanes and other users are unaffected.
Creations linked to previously flagged rug actors are blocked in the relevant lanes regardless of strategy risk level, alongside the standard scanner verdicts. In high-risk configurations those blocks become hard for automation while still informing manual users — the platform principle everywhere: automation gets the strictest standard because no human is watching.
It is not a profit machine and it does not remove market risk. It is discipline machinery: it applies the same gate to every decision, at 3 AM, without fear or euphoria. The strategy layer provides judgment; the safety stack provides boundaries; the operator provides the kill switch.
Next: operator analytics → All docs
Built for operators, not slippage goblins. Talk to us →