Make (Integromat)
Make lets you build automation sequences by dragging operations onto a canvas—no coding required. You're essentially replacing repetitive wo
Freemium
View tool Workflow automation tools connect the apps you use so data flows without a human copying and pasting. The AI layer changes the category in two meaningful ways: individual nodes now call LLMs to classify, summarize, decide, or generate, and the builder itself can turn a natural-language description into a working automation. The right tool depends on who is building (ops team vs. engineer), where the automations run (cloud vs. self-hosted), and how much logic you need beyond a linear trigger-action chain. Pricing models vary enormously across the category — per-task, per-step, per-operation, per-workflow — so forecast volume carefully before you commit, because the pricing model you pick will shape the shape of the automations you build.
Zapier is the right default for non-technical users and for teams where breadth of integrations is the binding constraint. The AI features (Chatbots, Agents, generative steps) are well-integrated into the builder, and the connector library is larger than any competitor — thousands of apps, most of them with real, maintained connectors rather than generic HTTP shims.
Make gives you a visual canvas, stronger branching and iterator logic than Zapier, and generally better pricing at medium volume. It is our pick for ops teams who have outgrown Zapier's linearity but do not want to self-host a full workflow engine. The visual scenario editor is particularly good at representing complex flows in a way that a new team member can actually read and change.
The workflow automation directory on this site is actively expanding. Self-hosted engines (n8n, Activepieces, Pipedream), specialist browser-automation tools, and AI-native agent builders will appear under the leaf categories as they are catalogued. For now, Zapier and Make cover the dominant share of real-world automation work.
Zapier if you want the widest connector library and the easiest onboarding. Make if you need more complex logic, better per-step cost, and a visual canvas that represents the flow clearly.
When your workflow has more than a dozen branches, when you need version control and tests, or when the cost of running at volume exceeds what the platform charges for self-hosted. At that point an engineering team with n8n, Airflow, or Temporal is cheaper and more reliable.
Use the platform's credential store, not inline environment variables. Rotate them, scope them narrowly, and audit which workflows use which credentials. Treat API keys with the same care you would database passwords.
Yes, with care. The platform becomes part of your uptime story, so pick a tool with a published SLA, retries, and clear observability. For latency-sensitive customer interactions, prefer direct API calls over a hop through a no-code platform.
Always validate the model's output against a schema before using it. If the validation fails, retry with a corrective prompt, route to a fallback branch, or send to human review — never blindly pass garbled output to the next step.
Good automations save meaningful hours per week and pay for themselves within a quarter. Bad automations consume more maintenance time than they save. Measure both sides — hours saved and hours spent maintaining — before celebrating a win.
Make lets you build automation sequences by dragging operations onto a canvas—no coding required. You're essentially replacing repetitive wo
