Home / By task / Automate a repeating task: trigger, connect, ship

Task guide · Automation

Automate a repeating task: trigger, connect, ship

If you only try one: Relevance AI

Why: Build a team of AI agents that share tools and hand off work

Ordered by checkable signals: link reachable → name. These candidates are AI tools, so there is no copy-paste install command and no star count. You can overrule it.

This page lists 8 automation and agent tools from our AI-tool database, ordered English UI first, then free tier, then source count.

See the workflow Compare the tools

8tools

0with a Chinese UI

0with an ICP filing

5with a free tier

First session: one chore, one trigger

Start with the chore you do every Monday.

Write the triggerPick stage.Wire the dataConnect stage.Run the failure drillPrompt 2.

01 / The process

Three stages: pick, connect, build

An automation you cannot describe the trigger for does not exist.

STEP 01

Pick the loop and the trigger

Write down one repeating task and what starts it. If you cannot name the trigger, no tool will help.

Deliverable: One loop with a named trigger

STEP 02

Connect tools and data

Pull the data from where it lives, then hand it on; keep one human approval where a mistake is expensive.

Deliverable: A run that touches real data

STEP 03

Build and host (advanced)

Only when the no-code route hits a wall: write the logic yourself, or rent a browser environment instead of maintaining one.

Deliverable: Something you host and can read logs from

Trigger first: a flow you cannot start on demand is not automated.

02 / Tool finder

Pick tools per stage

Filters only sort this page.

8 tools

Zapier

Pick

Stored one-liner: connects thousands of apps into automated workflows.

Evidence: site HTTP 200 · 1 source (curated:2026-10-22) · last checked 2026-11-03 · UI English · price tier free tier · ICP filing: not recorded

zapier.comOfficial site ↗

Make

Pick

Stored one-liner: visual builder for multi-step automations across apps.

Evidence: site HTTP 200 · 1 source (curated:2026-10-22) · last checked 2026-11-03 · UI English · price tier free tier · ICP filing: not recorded

make.comOfficial site ↗

Relay.app

Pick

Stored one-liner: workflow automation with human-in-the-loop approval steps.

Evidence: site HTTP 200 · 1 source (curated:2026-10-22) · last checked 2026-11-03 · UI English · price tier n/a · ICP filing: not recorded

relay.appOfficial site ↗

Lindy

Connect

Stored one-liner: build automations and email assistants by describing them in words.

Evidence: site HTTP 200 · 1 source (curated:2026-10-22) · last checked 2026-11-03 · UI English · price tier free tier · ICP filing: not recorded

lindy.aiOfficial site ↗

Relevance AI

Connect

Stored one-liner: build a team of AI agents that share tools and hand off work.

Evidence: site HTTP 200 · 1 source (curated:2026-10-22) · last checked 2026-11-03 · UI English · price tier n/a · ICP filing: not recorded

relevanceai.comOfficial site ↗

Browse AI

Connect

Stored one-liner: trains robots to scrape sites and watch pages for changes.

Evidence: site HTTP 200 · 1 source (curated:2026-10-22) · last checked 2026-11-03 · UI English · price tier free tier · ICP filing: not recorded

browse.aiOfficial site ↗

CrewAI

Build

Stored one-liner: open-source framework for multi-agent crews with roles and tasks.

Evidence: site HTTP 200 · 1 source (curated:2026-10-22) · last checked 2026-11-03 · UI English · price tier n/a · ICP filing: not recorded

crewai.comOfficial site ↗

Browserbase

Build

Stored one-liner: headless browser infrastructure for running web agents.

Evidence: site HTTP 200 · 1 source (curated:2026-10-22) · last checked 2026-11-03 · UI English · price tier free tier · ICP filing: not recorded

browserbase.comOfficial site ↗

Editorial note: entries come from our AI-tool database, picked by stage.

03 / Catalogue

One stored project

Same catalogue as the tool pages.

One catalogue project kept here.

awesomevivekvells/mcp-pandoc

Convert document formats between Markdown, HTML, PDF, DOCX, CSV and more using PandocMCP servers

587 ★

today

The four fields we can show: source link, last-seen date, README presence, public star count.

04 / Prompts you can paste

Two prompts we wrote

Plain text.

Both prompts are ours.

1. Turn a chore into a flowExample we wrote

This repeats every week: <fill>. It starts when: <fill>. Write the trigger, the three steps, and the one step where a human must approve.

Why: naming the trigger is half the work.

2. Failure drillExample we wrote

For this automation: <describe>. List what happens when step two fails, what should be retried, and what must page a human.

Why: automations break on the second step, not the first.

05 / New

New in the catalogue

The most recent additions in this category that pass our entry checks.

browsernode ↗🌐 Make websites accessible for AI agents. Automate tasks online with ease.2026-11-12

Reconmap ↗Reconmap is a collaboration-first security operations platform for infosec teams and MSSPs, enabling end‑to‑end engagement management, from reconnaissance through execution and reporting. With built-in command automation2026-11-12

skillsgate ↗Free, open-source desktop app to browse, install, edit, and organize AI agent skills across Claude Code, Cursor, Codex, and 21 more tools. macOS, Windows, Linux.2026-11-12

GeoAgent ↗A multimodal AI agent for geospatial data analysis and interactive visualization2026-11-09

clawdcursor ↗Local MCP server that gives any AI agent safe desktop control on Windows, macOS and Linux. Reads the screen as text and structure (accessibility + OCR) and acts by element name, with screenshots as the last resort. Multi2026-11-08

06 / Common questions

Before you begin

How many of these have a free tier?

5 of the 8 are recorded with a free tier; the other 3 are paid or unpriced.

Why this order?

English UI first (8 of 8), then free tier, then source count. 0 with a Chinese UI, 0 with an ICP filing.

Do I need to code to start?

完全不会写代码能开始吗？

What we actually check

Every label here - Chinese UI, ICP filing, free tier - comes from a field we recorded per tool. Where we found no evidence we write n/a rather than infer one. Pricing should be checked on the vendor page: what we show is what we saw on the day we looked.

Found something missing or wrong?

Tell us and we re-check it against the source before it changes.

Want a different job?

Task pages are grouped by what you are trying to finish.

Browse all tasks ↗

Data snapshot: 2026-11-14

← Digital human video · task 26 of 26 · 1 catalogue entries