Ideas branch. We build them.

Antler Labs

• OpenDriftliveOpenDrift helps you explore several possible answers to the same AI prompt. It collects, ranks, and saves responses with model-reported probability scores, making it easier to compare alternatives during evaluation, creative work, and experiments with language models.+

  OpenDrift explores the full probability landscape of an LLM response instead of locking you into a single sample. It sends a prompt through OpenRouter with Verbalized Sampling, asking the model to return several candidate answers with probability scores in structured XML.

  Those responses are parsed, ranked, and persisted so you can see where the model is confident, where it branches, and what alternatives it nearly chose: useful for evaluation, creative exploration, and understanding model uncertainty.

  Visit OpenDrift↗ (opens in a new tab)
• PayXorlivePayXor provides stablecoin payments and access control for Web3 applications. Prices are signed by your backend, and verified payments unlock the right features, sessions, or passes. A shared protocol, API, and SDK support integration across major EVM networks.+

  PayXor is on-chain payment infrastructure for Web3 apps: charge users in stablecoins, keep pricing backend-authoritative with EIP-712 signed quotes, and unlock entitlements from verified payments.

  One reusable protocol covers sessions, features, receipts, and passes across major EVM networks. A REST API and TypeScript SDK make integration straightforward; a transparent platform fee keeps economics simple for SaaS, games, marketplaces, and membership products.

  Visit PayXor↗ (opens in a new tab)
• LaCrewin developmentLaCrew brings budgets and approvals to organizations of AI agents. Each agent has an allowance and spending rules, with requests that exceed those limits moving up an approval chain. The aim is to let agents work while keeping control of the treasury with its owner.+

  LaCrew is the treasury and governance layer for AI agent organizations: payroll, budgets, and approvals so crews can spend within bounds without holding your root keys. Slogan in short: your agents, their budgets, your keys.

  Each agent sits in an onchain org chart with a smart account, a streaming allowance, and a policy stack that returns allow, escalate, or deny. Overages climb the tree like purchase orders; hiring, firing, and budget changes go through lightweight governance. Declared flows and connectors use the same vocabulary for side effects, while crew threads let agents plan, ask, and report with checkable receipts. Messages never authorize spend.

  The core protocol is open source and wallet-agnostic; the hosted cloud runs agents with scoped session keys so a compromise leaks bounded, expiring authority, never the treasury.

• FileOnChainin developmentFileOnChain packages documents, datasets, releases, and agent outputs into portable evidence that can be checked independently. It connects an artifact to its claims, signatures, and receipts, so verification can happen locally without depending on a hosted FileOnChain service.+

  FileOnChain turns digital artifacts (agent outputs, documents, releases, datasets) into portable evidence envelopes: JSON that binds digests to namespaced claims, signatures, and receipts from public storage and settlement systems. Anyone can verify locally and deterministically; no FileOnChain service is required in the loop.

  The stack separates a neutral evidence protocol, an opinionated Agent Evidence Profile for tamper-evident AI audit trails (hash-only by default), and FileOnChain Cloud for managed signing, settlement, API keys, and MCP. Open reference libraries ship under MIT so envelopes stay verifiable even after you leave the hosted product.

• Channlizein developmentChannlize connects a support widget on your website to the Slack channels your team already uses. Customer questions become threaded conversations, and replies appear back on the site. It is being built for teams that want to handle support within their existing workflow.+

  Channlize is Slack-native support: drop a React component on your site and every customer conversation lands in the Slack channel where your team already works, with no separate helpdesk to learn.

  Each inquiry becomes a threaded Slack conversation with routing by domain, page, or language; agents reply in-channel and the customer sees it in the widget in real time. A TypeScript React SDK, webhooks, email fallback when agents are offline, and operational metrics round out a platform built for small teams that refuse to context-switch.