Automations, described by what they do

15,351 working automation workflows. Every description is composed from the workflow's own steps: what starts it, which services it touches, whether it hands judgement to a model, and how much there is to read.

Browse the libraryBrowse by job

Cleaned of credentials and personal data. Ready to import.

Running now3 steps

Send daily Mailchimp subscriber reports to Slack

Waiting for a trigger
Workflows15,351working automations
Call a model7,615hand a step to AI
Services60apps they connect
Small5,286readable in one sitting

What they're for

17 kinds of job

The library

101 found

AllContentAlerts and monitoringData collectionIntegrationsLead generationResearch and reportingAI assistantsSchedulingMoney adminDocuments and filesChat assistantsCustomer supportE-commerceCRMHiringSetup and learningStrategy and planningQdrant ✕

Customer supportAIAnswer document questions with Ollama and QdrantStarts on someone opening a chat with it · 6 stepsQdrantCRMAILoading JSON with FTPStarts on an event in a connected app · 8 stepsFTPQdrantAI assistantsAIImplement on-prem RAG with Qdrant and OllamaStarts on someone opening a chat with it · 9 stepsQdrantDocuments and filesAIBuild a document-based AI chatbot with Google Drive and LlamaStarts on someone opening a chat with it · 11 stepsGoogle DriveQdrantAI assistantsAILocal chatbot with a local model and QdrantStarts on someone opening a chat with it · 11 stepsQdrantAI assistantsAIAI email autoresponder with OpenAI and GmailStarts on an arriving email · 12 stepsEmailGmailQdrantAI assistantsAIAI-powered email processing autoresponderStarts on an arriving email · 12 stepsEmailGmailQdrantAI assistantsAIAI-powered email processing autoresponderStarts on an arriving email · 12 stepsEmailGmailQdrantAI assistantsAIBuild a RAG agent with Qdrant and OpenAIStarts on someone opening a chat with it · 12 stepsQdrantAI assistantsAIBuild a ServiceNow knowledge chatbot with OpenAIStarts on someone opening a chat with it · 12 stepsQdrantSchedulingAIClassify and route DevOps chat requests with MattermostStarts on another system calling it · 12 stepsGoogle CalendarQdrantMoney adminAIGenerate AI-curated sales quotes with OpenAI and QdrantStarts on someone submitting a form · 12 stepsEmailQdrantAI assistantsAIBuild persistent chat memory with OpenAI and QdrantStarts on someone opening a chat with it · 13 stepsQdrantChat assistantsAICreate an AI Telegram bot with Google Drive and QdrantStarts on an incoming Telegram message · 13 stepsGoogle DriveQdrantTelegramData collectionAIMatch medical symptoms to products with OpenAI and QdrantStarts on someone opening a chat with it · 13 stepsGoogle SheetsQdrantResearch and reportingAIAI crew to automate fundamental stock analysis - Q&A workflowStarts on someone opening a chat with it · 14 stepsGoogle DriveQdrantAlerts and monitoringAIAnalyze alerts from Alertmanager and send diagnostic reportsStarts on another system calling it · 14 stepsQdrantSlackMoney adminAIPDF proposal knowledge base with OpenAI and Qdrant RAGStarts on someone opening a chat with it · 14 stepsQdrantS3AI assistantsAIChat with Gemini and QdrantStarts on someone opening a chat with it · 15 stepsQdrantCustomer supportAIClassify and route DevOps support messages with SlackStarts on another system calling it · 15 stepsQdrantSlackAI assistantsAIConvert DevOps Mattermost requests into Jira tasksStarts on an event in a connected app · 15 stepsJiraMattermostQdrantAI assistantsAIEvaluate tool usage accuracy in multi-agent AI workflowsStarts on someone opening a chat with it · 15 stepsQdrantChat assistantsAIInvestigate CI/CD incidents in Slack with OpenAIStarts on an event in a connected app · 15 stepsQdrantSlackCustomer supportAIAnswer infrastructure questions in Slack with OpenAIStarts on an event in a connected app · 16 stepsQdrantSlack

Start small

A few you can read in one sitting

What we took out

Cleaned before anything was stored

These files carry more of whoever built them than is obvious. All of it is stripped, and every file was checked, not sampled. Where an author left a live document behind, it is replaced with the placeholder the careful ones used, so you can still see what belongs there.

Triggers

What starts them

  1. A clock4,166
  2. An event in a connected app3,946
  3. Another system calling it2,801
  4. An arriving email1,591
  5. Someone submitting a form1,545
  6. Someone opening a chat with it895
  7. A change in a Google Sheet216

Services

What they touch

Google Sheets5,229Gmail3,419Slack2,307Telegram1,809Google Drive1,540Email863Airtable554Google Calendar480Notion471RSS404Postgres306WhatsApp285

Read the workflows

Filter 15,351 automations by what they do, what they connect to and how big they are, and open the definition itself, ready to import.

Q&A

What are these?

Working automations: the definition of each, ready to import, with a description composed from its own steps.

Why are some descriptions unflattering?

Because they say what the workflow does. One that is four steps and a webhook is described as four steps and a webhook, so you can decide from the page whether it is worth your afternoon.

What was removed from the files?

Instance identifiers, canvas notes with personal links and email addresses, saved credential names, and live documents left in by their authors. Every file was checked, not sampled.

What do I need to run them?

A workflow automation tool that imports these definitions. Each imports as it is; you add your own credentials.