What is a Taskade automation?
A Taskade automation runs steps for you when something happens. A trigger starts it, actions do the work, and your projects update by themselves.
TL;DR: Automations are the Execution layer of Workspace DNA. A trigger pulls an event in, control steps route it, AI steps think, and actions push data out across 100+ integrations. Automations run on every plan, Free included. After 5 failed runs in a row, an automation turns off and owners get an email.
How do you build an automation?
You build an automation in one of three ways, and each way ends in the same editor with the Flow, Form, and Runs tabs.
- Ask Taskade EVE. Pick the Automations mode and describe the job in plain English. See Workflow Generator.
- Start from a template. Open Quick Start and pick a ready automation. See Automation Prompts.
- Build by hand. Add a trigger, then add each step in the Flow tab.
Which steps can a Taskade automation use?
A Taskade automation combines triggers, control flow, AI steps, and actions. This table lists each part with its exact label and plan.
| Part | What it does | Labels in the editor | Plan |
|---|---|---|---|
| Taskade triggers | Start from work in Taskade | When a task is assigned, When a task is due, When a field changes, When an email arrives, When an agent chat ends, When you run this flow, and more | Every plan |
| Schedule | Run on a timetable | Every hour, Every day, Every week, Every month, Custom schedule | Every plan |
| Fast schedule | Run every few minutes | Every few minutes | Any paid plan |
| Form | A shareable form starts the automation | Form tab | Every plan |
| Incoming webhook | Another app starts the automation | Webhook Received | Pro and above |
| Branch | Take a different path by condition | Choose a path | Every plan |
| Filter | Stop unless a condition is true | Only continue if | Every plan |
| Loop | Repeat steps once for each item in a list | Repeat for each | Every plan |
| Delay | Wait, then continue | Wait, then continue: Delay For, Delay Until | Every plan |
| AI steps | Write, summarize, sort, or reply with AI | Ask AI, Write with AI, Summarize with AI, Categorize with AI, Reply with AI | Every plan, uses credits |
| Agent steps | Hand the work to an AI agent or a team | Ask an agent, Ask an agent team, Run an agent command | Every plan, uses credits |
| Integration actions | Push data out to another app | One group for each app | Every plan. PRO connectors need a paid plan |
| Web and MCP | Call any API or a remote MCP server | Send HTTP Request, MCP Client | Every plan |
A filter that stops a run marks it Filtered, not failed. Ask AI can return plain text or structured output that follows a JSON schema.
Pick a model for an AI step
Each AI step has a model menu. Pick the tier by the job, not by the model name.
| Tier | Use it for | Plan |
|---|---|---|
| TSK-Auto | Taskade picks the best model for each step. This is the default | Every plan |
| TSK-Instant | Short, simple steps with the lowest credit use | Every plan |
| TSK-Standard | Everyday writing, sorting, and replies | Any paid plan |
| TSK-Thinking | Bigger steps that need more care. Uses more credits than TSK-Standard | Any paid plan |
| TSK-Reasoning | Complex steps that need deep thinking. Uses more credits | Business and up |
What happens when an automation run fails?
When a run fails, Taskade tells you, and a broken automation stops by itself. You then fix it and turn it back on.
- Notices. Owners get a notice after the 1st and the 3rd failed run in a row.
- Automatic pause. After 5 failed runs in a row, the automation turns off and owners get an email.
- Reset. A successful run resets the count. A filtered run does not count as a failure.
- Run history. Open the Runs tab. Filter by All, Completed, Failed, or Running.
Each plan includes a number of runs for each billing period. The whole workspace shares them. They are not counted for each person.
| Plan | Automation runs for each billing period |
|---|---|
| Free | 10 |
| Pro | 1,000 |
| Business, Max, Enterprise | 10,000 |