Image source: Rerun official homepage. Screenshot captured November 7, 2026 (Vietnam time), for editorial identification.
Rerun is a platform for assigning recurring work to AI coworkers that operate in a cloud workspace. Its appeal is practical for small teams: define a job, connect the necessary tools and decide when it should run. The recent news is a relaunch and product update, rather than the first appearance of the service.
Following its November 5, 2026 Product Hunt launch, Rerun published version 0.10.0 on November 6. That release adds workspace roles and more precise controls over access to Boxes, its work-organizing areas. This overview uses public documentation and pricing, not hands-on testing.
What can Rerun coworkers do?
The official website describes coworkers with their own cloud computer, connected apps, retained context and scheduled or triggered tasks. Example roles include research, reporting and administrative work. Rerun says it supports more than 3,000 integrations and allows users to connect supported AI subscriptions or API keys, or use Rerun credits.
For a team already copying AI output into several business tools, this may reduce manual handoffs. It also changes the risk: an assistant with app access can affect real work. Rerun says consequential actions pause for approval, including spending money, sending client email and deleting information. Those controls should be tested before depending on unattended runs.
What changed in the November update?
Version 0.10.0 separates Admin and Member responsibilities. It also lets a Box be shared across a workspace or restricted to named people, with use or manage rights. The release notes describe these controls across the web app, workspace assistant and iOS.
That distinction matters for teams serving multiple clients. A person who needs one project’s output should not automatically need authority over provider connections or every other project. The upgrade notes say every existing member becomes Admin and every Box remains open until an administrator restricts access. Existing users should review these defaults and deliberately set their intended boundaries; upgrading alone does not make shared work private.
A hypothetical weekly reporting workflow
Imagine a small agency that prepares a weekly summary from a campaign spreadsheet and approved notes. A sensible first assignment would be to produce a draft report for review, with no permission to change campaigns or send messages externally.
- Choose one reporting period, one data source and a clear output format. Define which numbers should be included and how they should be calculated.
- Connect only the information needed for that job. Keep unrelated customer records and payment tools outside the initial setup.
- Run the task manually and compare the output with the source spreadsheet. Check totals, date boundaries and missing rows.
- Ask the coworker to flag missing information instead of estimating it. Make the human reviewer responsible for accepting the final report.
- Only then schedule another run. Recheck the result after a source format, account permission or reporting requirement changes.
This example is an evaluation plan, not a reported customer outcome. The useful measurement is whether the report becomes faster to verify without introducing new errors, rather than how many tasks the agent appears to complete.
Rerun pricing
The current official pricing page lists Pro at $29 per month, with unlimited coworkers, unlimited seats and 500 monthly credits. A five-day trial requires a card and includes up to three coworkers and 350 credits. Canceling before the trial ends avoids the charge, according to the page.
LinkedIn and WhatsApp access is listed as a $19-per-month add-on, while an additional 1,000 monthly Rerun credits costs $10. Connecting an existing model plan does not remove that provider’s limits. Check the selected billing period, provider costs and any add-ons before subscribing.
Who should consider it?
Rerun is most relevant to teams with repeated, well-defined workflows and someone available to review exceptions. It is a less comfortable fit when requirements change constantly or nobody can check whether an automated action was appropriate.
Review permissions, credential handling and data retention before connecting business accounts. Start with drafting and read-only work, then expand only when the results and approval behavior are understood. The combination of recurring tasks and shared access can be useful, but it deserves the same care as granting a new teammate access to company systems. Compare related products in GPTOnline’s AI tools directory.
