| Who each product is built for | Work on Repeat is built for solo businesses, consultancies, startups, and lean teams putting recurring knowledge work on a schedule. | Relevance AI presents agents and AI workforces for coordinating recurring or queued tasks; its Tasks documentation defines a Task as one run of work carried out by an Agent or a Workforce. |
|---|
| How work is defined | A routine owns its instructions, model, inputs, schedule or API trigger, connected tools, and delivery settings. | Agents receive Tasks made up of messages, steps, and Tool runs, while a Workforce Task View monitors and manages work across agents, including concurrent tasks, approvals, and human intervention. |
|---|
| How recurring work starts | Routines support five-field cron schedules with an IANA timezone, plus manual and API runs. | Relevance AI documents daily, weekly, monthly, and custom recurring schedules, six-field AWS EventBridge cron expressions, multiple recurring schedules for one agent, and future scheduled messages. Its scheduling product page also documents work hours, workload pacing, bulk scheduling, wake mode, and a queue manager; the current pricing comparison lists Work Hour Controls specifically for Enterprise. |
|---|
| How systems are connected | Built-in providers plus generic remote MCP connections expose approved tools to routines. | Relevance AI documents native integrations with ready-made actions and triggers, plus Pipedream integrations extending its catalog to 2,000+ services. Its MCP client connects agents to preset remote MCP services or a custom server URL, can connect multiple instances independently, and does not currently support local JSON-configured MCP servers. |
|---|
| How actions are bounded | Every MCP tool begins disabled; enabling it and allowing automatic execution are explicit per-routine decisions. | Workforce edges can use Auto Run, Approval Required, or Let Agent Decide modes; Auto Run can be capped with Max auto runs, while approval requests and escalations can be reviewed through Workforce Task View. |
|---|
| How runs and usage are inspected | Work on Repeat persists run, model-step, tool-call, error, usage, and webhook-attempt history. | Relevance AI's Tasks surfaces show task lists and queues, messages and Tool runs, search and filters, status, credits used, run time, and linked tools across Agents and Workforces. Workforce Task View documents conversation and task logs, decision points, tool usage, task timelines, approvals, failures, and concurrent work; the current pricing table lists 90-day Task History for Pro and Team. |
|---|
| How packaging is measured | Free is $0 with no card and includes 1 active routine, 10 run starts per UTC month, 1 connection, 7 days of history, 1 concurrent run, 3 model steps per run, read-only API access, and $0.50 of upstream AI usage. Pro is $39 monthly and includes 10 active routines, 250 run starts per Stripe billing month, 10 connections, 90 days of history, 2 concurrent runs, 4,000 model steps per run, full API access, signed result webhooks, and $10 of upstream AI usage. Business is agreement-based, with allowances configured for the customer's operating requirements. | As reviewed on 2026-08-13, Relevance AI lists Pro at $19 per month on annual billing or $29 month-to-month, with 30,000 Actions per year or 2,500 per month, $240 in Vendor Credits per year or $20 per month, Schedule Tasks, Bring Your Own LLM, and 90-day Task History. Relevance AI defines an Action as an agent running a tool, which may be a simple operation or a complex workflow with many steps; Vendor Credits cover LLM and tool usage. Neither meter is treated as a Work on Repeat run start or model step. |
|---|
| What switching requires | Users manually recreate one responsibility, connect the least tools needed, prove a manual run, inspect the result, and then schedule it. | The official Relevance AI sources reviewed for this profile do not document an automatic import into Work on Repeat, so none is promised here. |
|---|