Rerun

Product Introduction

1. Definition: Rerun is an open-source (AGPL v3) AI coworker platform and multi-agent orchestration workspace that converts existing LLM subscriptions—such as ChatGPT Plus/Pro and Claude Pro/Max—or API keys into autonomous AI employees. Categorized under AI workflow automation and agentic software, it supplies dedicated cloud-hosted virtual machines, native computer-use capabilities, and more than 3,000 app integrations to execute complex, multi-step business operations.

2. Core Value Proposition: Rerun bridges the gap between static chatbot answers and automated business execution by employing persistent AI agents that operate independently on scheduled intervals or webhooks. By leveraging users' existing AI subscriptions (ChatGPT, Claude, Grok, or OpenRouter), it drastically reduces API inference costs while providing a secure, human-in-the-loop environment to handle sales prospecting, SEO audits, marketing analytics, and administrative overhead without manual copy-pasting.

Main Features

1. Bring Your Own Model (BYOM) & Subscription Integration: Rerun allows users to authenticate directly via their existing ChatGPT, Claude, or Grok accounts, or utilize API keys from Anthropic, OpenAI, Google Gemini, and OpenRouter. This eliminates redundant token expenses while allowing individual agent assignments based on task complexity (e.g., routing simple inbox sorting to economical models and complex reasoning to frontier models).

2. Autonomous Computer Use & 3,000+ Integrations: Each AI coworker operates on a dedicated cloud machine with its own storage, memory, and interface access. Beyond pre-built API connectors for enterprise tools like Stripe, HubSpot, Notion, and Google Drive, agents feature native computer-use capabilities to interact with legacy web applications, click through software dashboards, and execute tasks even in the absence of native integrations.

3. Scheduled Routines, Webhooks, & Multi-Agent Collaboration: The platform features a robust workflow scheduler and event-driven trigger system capable of executing tasks on custom cron schedules (e.g., daily at 7:00 AM) or via external webhooks. Coworkers can be chained together sequentially—such as an agent sourcing leads passing data to an outreach agent, which then forwards final documentation to a CRM management agent—all housed within organized workspaces called "Boxes."

Problems Solved

1. Pain Point: Traditional generative AI interfaces function purely as conversational query-response bots, requiring human operators to manually copy text, toggle between browser tabs, and log into disparate third-party portals to execute workflows. Furthermore, chatbots reset their context daily and lack persistent memory of business-specific nuances like client payment terms or historical project scopes.

2. Target Audience: Digital marketing agencies, early-stage startups, e-commerce managers, real estate firms, SaaS founders, and freelance professionals seeking to scale operational capacity without scaling headcount.

3. Use Cases: Automated financial invoice chasing and payment tracking, weekly SEO site audits and competitor ranking reports, cross-channel paid ad spend analysis, continuous social media content planning and publication, and end-to-end sales lead generation and scoring.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros: Zero marginal inference costs for users leveraging their existing ChatGPT Pro or Claude Max subscriptions; open-source and self-hostable (AGPL-3.0); human-in-the-loop safety rails that pause execution before high-risk actions (e.g., sending emails, spending funds); persistent cloud memory that retains cross-session business context; and support for team collaboration with zero per-user pricing fees.
   • Cons: Reliance on the operational uptime of third-party consumer LLM interfaces when using account-linking features; initial setup requires mapping tool integrations and defining precise execution boundaries; complex multi-app web navigation can occasionally necessitate manual human intervention if UI layouts unexpectedly change.

2. Key Alternatives & Differentiation:

   • Rerun vs. Custom Agent Frameworks (e.g., AutoGPT, CrewAI): While custom frameworks require extensive Python development, API management, and self-hosted infrastructure, Rerun provides a turnkey SaaS-like collaborative workspace with built-in GUI management, subscription token pass-through, and ready-made business templates.
   • Rerun vs. Standard Automation Tools (e.g., Zapier Central, Make): Traditional integration platforms rely on rigid, deterministic "if-this-then-that" logic. Rerun introduces cognitive reasoning, computer-use UI navigation, and unstructured decision-making capabilities, allowing agents to handle ambiguous inputs like drafting contextual replies or auditing unstructured data.

Frequently Asked Questions (FAQ)

1. How does Rerun utilize my existing ChatGPT or Claude subscription? Rerun securely connects to your active ChatGPT or Claude account via session authentication, allowing your AI coworkers to execute tasks using the credits and model intelligence you already pay for, thereby minimizing additional token billing costs.

2. What happens if a coworker attempts a risky action like sending an email or spending money? Rerun enforces strict human-in-the-loop safety protocols. Before executing any consequential or high-risk action—such as financial transactions, client communications, or data deletion—the AI coworker automatically pauses and requests explicit approval from a team member via the dashboard or email notification.

3. Does Rerun continue running when my personal computer is turned off? Yes. All AI coworkers operate on dedicated, isolated cloud virtual machines hosted in Europe that run 24/7. Scheduled routines (such as 7:00 AM invoice chasers or Monday morning analytics reports) execute continuously regardless of whether your local machine is shut down.

4. Is my business data secure and private? Yes. Your workspace runs on an isolated private machine walled off from other tenants. Rerun does not train its models on your proprietary business data, and underlying model providers (like Anthropic and OpenAI) are contractually prohibited from utilizing API data or linked session payloads for model training.