Rerun is an AI-powered work automation tool built around “Capable Coworkers”: agents that take on a job, use a cloud computer and connected business apps, and carry out the steps rather than simply suggesting what a person should do. Users describe a task in plain language or choose a ready-made coworker, connect their own AI subscription or API key, and can let work run on a schedule or in response to an event. The product is aimed at founders, freelancers and small teams looking to automate recurring operational work.
Key features
Agents that work across apps
Rerun says its coworkers can use a browser, files and a terminal on their own computer, with a screen the user can watch and take over. The page lists more than 200 app connections, including Gmail, Slack, Notion, Stripe and HubSpot. A Rerun assistant can help set up coworkers around a stated goal. This makes the product more task-oriented than a chat interface that returns text for someone to copy and act on.
The page also describes lead enrichment using a connected Apollo, Clay or Prospeo account. Enrichment can include verified email addresses, phone numbers and company information. LinkedIn and WhatsApp messaging are separate paid add-ons, not included functionality.
Scheduled and event-triggered work
Coworkers can run on a schedule or start when an event occurs, such as a new email, a Stripe event or a webhook. Their computer runs in the cloud, so work can continue when the user's laptop is closed. Results can be delivered in the chat, a file or Slack, along with a record of what the coworker did and where it became stuck.
The product includes approval checkpoints for actions such as sending, paying or deleting. The user can approve an action once, for the current session, or always. If a coworker needs input, a team member can respond so it can continue from where it stopped. The activity view shows the steps taken, the app called and the response received.
Shared workspace knowledge
Rerun provides a shared company memory for information such as client details, prices and tone, with sources attached so coworkers have context. Successful processes can be turned into skills that people can read and edit; the coworker asks before changing a skill. The page says coworkers, team seats, memory and skills are shared at workspace level, rather than being tied to an individual who might leave.
Who it is for
Rerun is best suited to small businesses with repeatable tasks involving several apps, such as sorting an inbox, following up on invoices, finding leads or preparing a regular report. It may also suit teams that want to describe a desired outcome in everyday language instead of constructing a fixed sequence of automation steps. The product page says coding is not required and that users can start from a ready-made coworker.
It is less suited to someone looking only for a conversational AI assistant or a tool that supplies AI usage itself. Rerun requires a separate AI subscription or API key from a supported provider. The product page says the available options include Anthropic, OpenAI, Google and OpenRouter, and also notes that custom endpoints are supported.
Limitations
AI usage is not included: users connect their own subscription or API key, so the cost and setup of that AI service remain separate. Lead enrichment also relies on the user's own Apollo, Clay or Prospeo account. LinkedIn and WhatsApp messaging are not included and require separate add-ons.
Rerun is described as an early-stage product, and the page says the team is still building it. While it identifies a dedicated cloud computer hosted in Europe, it does not provide detailed answers on the page about matters such as encryption, staff access or infrastructure capacity. Teams assessing sensitive workflows may want those specifics before relying on the service.
Verdict
Rerun combines AI agents, app connections, cloud execution, approvals and shared business context in a workflow tool for teams that want recurring jobs handled with less manual intervention. Its distinguishing approach is to give coworkers a computer and a goal, while keeping people in the loop for consequential actions. The main trade-offs are dependence on a separately supplied AI service, additional requirements for lead enrichment and messaging, and the limited operational detail available for an early-stage product. It is most relevant to small teams that can provide their own AI access and want to automate practical, multi-step business tasks.


