Town vs OpenAI Dots: Saved Routines or Adaptive Delegation?

Town is a useful candidate when you can define a responsibility as a routine with clear triggers and tools. OpenAI dots is a useful candidate when one ongoing coordinator needs to juggle priorities and delegate separate tasks. Both can support recurring work; compare how the work is expressed, inspected and stopped.

Official documentation checked October 23, 2026. Agentastic authored this guide. The workflow below is a suggested comparison, not a performance test.

Make the saved responsibility visible

Town’s routine guide and modes explain its configuration. Dots’ tasks/memory and controls explain ongoing coordination. These are different interfaces for responsibility, not evidence of a universal winner.

Follow an issue from report to weekly update

Suppose a team wants new issue reports investigated and a Friday summary of unresolved work. The first part is event-driven. The second part is scheduled. Sending a summary to the owner is also different from changing the issue status or messaging a customer.

For Town, configure the relevant trigger, connected tool and routine mode. Its documentation makes that combination explicit. A read-only routine can report to you while blocking source modifications, which may fit the initial investigation stage. If you need live events, inspect the plan rather than assuming a daily schedule provides equivalent timing.

For a dot, ask it to confirm the event it can monitor and inspect any delegated tasks it starts. OpenAI says connecting a service alone does not create monitoring. Separate delegated threads let you inspect the work while the dot continues coordinating. If investigation depends on local files, the computer connection becomes part of the requirement.

A useful evaluation follows one issue through a status change. Does Friday’s summary distinguish a resolved report from a still-open decision? Can you identify why the agent resumed work? These are evaluation questions, not results we have measured.

Messaging access does not settle execution

Town documents shared context across web, email and messaging channels. OpenAI likewise documents continuity with the same dot across supported contact methods. Neither fact alone grants source-account or computer permissions.

Dots can connect one personal computer separately; local work needs it online with the ChatGPT app open. The cloud browser has its own sessions. Computer connection guide. Town’s underlying local-versus-cloud execution architecture is not established by the pages reviewed, so its desktop client is not evidence of equivalent local access.

Allowance and cancellation can decide the fit

Town’s plans gate event/calendar triggers above Starter’s limited daily schedules; paid usage consumes credits. Starter is $15 billed monthly, while the listed Pro $49 monthly rate assumes annual billing. For dots, eligible-plan rollout and delegated Work/Codex allowances matter; Enterprise must be enabled by an administrator. Dots overview, admin guide.

OpenAI explicitly separates pausing the dot from stopping delegated work and canceling schedules. For the issue workflow, confirm that both event handling and Friday’s report are stopped when ending the responsibility. Do not assume Town’s routine controls share identical pause semantics.

Choose Town to evaluate explicit routine configuration and trigger tiers. Choose dots to evaluate an ongoing coordinator with visible delegated work and documented connected environments. For a shared Bot workspace, see Grok Bot vs Town, or use the 21-pair matrix.