Loomlay
Launch agentic trading strategies in one-click
About
The problem it solves Web3 traders need fast, reliable, and low-cost automation to seize rapid trading opportunities. Current tools have limitations: Manual Trading: Too slow, emotionally driven, error-prone, can't run 24/7. Generic Bots: Complex to set up, rigid, costly due to inefficient polling and delays. Previous Loomlay Agents: Relied on slow and expensive repeated checks or social triggers, leading to poor performance and high cost. Also limited to one agent per user, restricting shared strategy availability. Our new "Smart Events" engine solves these key issues directly by: Reducing Latency & Costs: Agents now instantly react to clearly defined crypto events (such as price thresholds, trading metrics, new social media posts, technical indicator change), eliminating slow polling and high AI costs. Simplifying User Experience & accessibility: Users easily set events and trading logic using plain English or choose ready-made strategies from our store. Built-in secure wallets (powered by Turnkey) provide safe, hassle-free implementation and sharing of strategies. Expanding Strategic Possibilities: Supports highly customized event-driven trading like conditional buys/sells, dynamic dollar-cost-averaging, monitoring influential wallet moves, significant token-holder changes, or social media alerts from influential accounts. Loomlay Smart Events solution delivers powerful, intuitive, and affordable AI-driven crypto trading, accessible to a broader audience and suitable for sophisticated strategic requirements.
How it's made
Challenges we ran into Wallet Delegation: We engineered a system using Turnkey for secure, non-custodial agent wallet management, ensuring users grant permissions without exposing private keys or allowing agent creators to misuse funds. This included making workflows transparent and planning re-delegation safeguards. Universal & Complex Event Processing: A core challenge was building an engine to interpret and act on a vast array of arbitrary user-defined event triggers, often involving complex, nested conditional logic (e.g., "if price X AND volume Y OR buyers amount increases by A percent"). This demanded a highly flexible database structure and a robust real-time evaluation system for both on-chain and off-chain data. Natural Language to Structured Commands using LLMs: To ensure user-friendliness, we tackled the complex task of accurately translating plain English instructions (e.g., "if 4-hour trade count > 50, buy $1 of X") into precise, machine-readable, structured event definitions for our backend engine. This required integrating Large Language Models (LLMs) and developing specific parsing logic to map free-form text to our internal event schema. Real-time, Scalable Event Ingestion: Ensuring agents reacted instantly required processing of high-volume event data from diverse sources (blockchain, APIs, social media), crucial for timely trade execution. These required a multi-faceted approach, blending secure infrastructure, sophisticated database design, advanced LLM integration, and a scalable event-driven architecture.