AIAny
AI Agent2026
Icon for item

Harvey

Runs a local sales workflow that finds prospects, drafts outreach, and handles replies using a Claude Code or Codex subscription instead of a separate model API bill. Outgoing emails wait for approval by default.

Introduction

Subscription-based model access changes the economics of an always-on sales agent, but it does not make outreach risk-free or hands-off. Harvey uses Claude Code or Codex for judgment-heavy tasks while ordinary pipeline decisions and sending checks are handled by deterministic code. That division is the key idea: reserve model calls for personalization and conversation, not for work a program can decide directly.

What Sets It Apart
  • A complete prospect-to-reply loop: it discovers businesses, profiles their websites, enriches and verifies contact details, drafts sequences, and classifies replies. This connects research to follow-up instead of stopping at a generated email.
  • Human review is the default: messages enter an approval queue, and a deterministic pre-send gate checks items such as recipient validity, merge fields, and message rules. This makes it better suited to supervised outreach than an agent that sends freely.
  • The model is only one component: user-confirmed business signals define targeting, while collection, scheduling, limits, and pipeline state are managed locally. The result is more inspectable than delegating the whole workflow to a model.
Great Fit If…

It suits small teams that sell to local businesses, want a locally stored prospecting pipeline, and already have an eligible Claude or ChatGPT subscription plus a dedicated sending mailbox. Look elsewhere if you need a managed sales platform, broad enterprise contact-data coverage, or outreach without configuring email infrastructure and reviewing legal, consent, and deliverability requirements. Subscription usage limits still apply, and automated cold email remains the user's responsibility.

Information

Categories

More Items

Hugging Face
AI Model2026

Coordinates long-horizon scientific and engineering tasks by delegating independent investigations to parallel subagents and revising its approach through evidence checks. It combines tool use and multimodal input in a 35B mixture-of-experts model.

GitHub
AI Agent2026

Shows what fills a DeepSeek Harness agent's context and how it changes per request. Browse prompts, tools and messages, trace compaction, and inspect token usage, estimated costs and parent/subagent relationships.

GitHub
AI Coding2026

Adds file editing, Git views, task monitoring and side chats to DeepSeek Harness. An extensible sidebar and split bottom workbench keep files, terminals and agent activity beside the conversation.