Web API automation

2016-06-04 — 2025-11-10

Wherein the art of web API automation is described as a system of persistent change‑checking services and numerous prebuilt connectors, and is noted to be implemented via webhooks and hosted functions.

browser
compsci
computers are awful together
diy
doing internet
faster pussycat
Figure 1

Like browser automation but without the browser, web API automation tools use the published access points of various web services (Dropbox, Facebook, GitHub, etc.) to automate stuff for us.

For a while, automations like these were how we thought the internet would be run — a smooth, integrated flow of information for users. There’s been a lot less hype about that recently. Cynically, this might be because the best way to monetize users is not to make them more efficient but to make them slower so they will look at more advertising or watch more conspiracy videos etc. Despite the claim that APIs are dead, these web APIs are still a going concern.

A lot of them are examples of low code development adapted to a particular domain, or applications of graphical flow-based programming. What do all these services provide that’s hard to replicate by just running a script on my laptop? As far as I can tell, it’s that they

  1. generally make it easy to run a persistent service that checks to see if things have changed (e.g. new email, new tweet, new file in Dropbox)
  2. provide a large library of pre-built connectors to web services and often get those services to actively participate in the automation (e.g. webhooks)
  3. mitigate automation hell

All those features sound high value to me, and yet in practice I don’t use them; I vibe-code my own automations instead. Occasionally I use lightly hosted “serverless” functions via Fly.io or runpod; once again vibe-coding them. I don’t know if my needs are idiosyncratic, but I’ve always ended up needing something custom that these platforms don’t support well: weird filtering, a badly supported service, or an odd schedule…

TODO: discuss agentic AI doing automation.

1 Commercial Automation Services

  • IFTTT: The OG consumer-focused, no-code platform that connects apps and devices to “automate business & home”. Its value proposition is its simplicity. It’s built on “Applets,” which are simple conditional statements: “If This, Then That.”
  • Zapier: A “leader in easy automation” primarily for business and “prosumer” users (relative to, say, IFTTT). Its value is its massive library of over 7,000 app integrations, empowering non-technical users to build complex, multi-step automated workflows (“Zaps”) that move data and automate processes between their web applications.
  • Val Town: A “social website to code in the cloud”. Its value proposition is being a “Zapier for developers” or “AWS Lambda that’s fun”. It eliminates all setup and configuration, allowing you to write and run TypeScript functions (called “vals”) as APIs, scheduled tasks, or email handlers directly from the browser in seconds.
  • Botize: An automation platform whose value proposition is ease of use. It positions itself as an alternative to complex visual editors by allowing users to “Describe what you need” in plain language. The service then attempts to build the required automation from that description.
  • Pipedream: An integration platform built specifically for developers. Its value is providing a serverless infrastructure to run “workflows” as code (Node.js or Python). It offers managed authentication for thousands of APIs, data stores, and the ability to build and run complex integrations that require custom logic beyond what no-code tools can offer.
  • Microsoft Power Automate: An enterprise automation platform. Its primary value proposition is its deep integration into the Microsoft 365 ecosystem (SharePoint, Teams, Excel, etc.) and Azure. It’s designed for “citizen developers” within large organizations, offering robust low-code automation with strong governance, security, and AI (Copilot) capabilities.
  • n8n.io: A “flexible AI workflow automation platform for technical teams”. Its key value proposition is its “fair-code” model, which allows it to be self-hosted (on-premises or in a private cloud). This gives companies full control over their data, ensuring privacy and compliance. It is a “node-based” visual tool that also allows developers to extend any part of it with custom JavaScript or Python code.

2 Self-hosted Open Source Automation services

  • n8n-io/n8n (GitHub): (See above) This is the source code repository for the n8n platform, which emphasizes its self-hostable nature for data privacy and its flexible node-based system that combines low-code and pro-code.
  • Automatisch / Automatisch (GitHub): An “open-source Zapier alternative”. Its value proposition is built on two main points: it is self-hostable, giving users complete data privacy (critical for GDPR or sensitive industries), and it is open-source (AGPL-3.0 license), which means no vendor lock-in.
  • Huginn: A “hackable version of IFTTT or Zapier on your own server”. This self-hosted, open-source system allows you to build “agents” that monitor the web, watch for events, and take actions. Its value proposition is its power in web scraping and monitoring, allowing you to create custom event feeds, track website changes, or receive digests on any topic.
  • nodered: A “low-code programming tool for event-driven applications”. Its value is its visual, flow-based programming interface (where you “wire” nodes together) that is built on Node.js. It is lightweight and ideal for running on low-cost hardware like a Raspberry Pi, making it a dominant tool in the Internet of Things (IoT) and home automation space for connecting hardware, APIs, and online services.

3 Hosted Functions / Serverless

  • Fly.io: A platform for developers to deploy applications and databases globally. Its value proposition is running apps “at the edge,” physically close to users, to dramatically reduce latency. It works by turning standard Docker containers into lightweight virtual machines that can be launched quickly across its 35+ global regions.
  • runpod: A serverless cloud platform specifically designed for AI and machine learning workloads. Its value proposition is providing on-demand, auto-scaling GPU compute. This allows developers to run compute-intensive AI models without managing servers, paying only for the processing time they use. That’s often significantly cheaper and faster than traditional cloud providers.