AI automation tools,
and how to choose them.

The two kinds of tools

Most AI automations are built from two kinds of tools working together. If you are new to the idea, start with our guide to what AI automation is.

• Workflow platforms such as n8n, Make, and Zapier connect your apps and move data between them. They decide when something happens and where the result goes.
• AI models from providers such as OpenAI, Anthropic (Claude), and Google (Gemini) do the understanding: reading a document, summarising a message, or classifying a request.

Around them sit the everyday tools that do the actual work, such as calendars, email, SMS, and payments. Choosing well means picking the right tool for each layer, not one tool for everything.

n8n vs Make vs Zapier

All three let you build automations without writing much code. They differ in control, flexibility, and how they grow with you.

Plans and limits change often, so check each provider's current pricing before you decide.

Our rule of thumb: Zapier for quick wins, Make when a visual builder suits your team, and n8n when the workflow is business-critical, needs custom logic, or should run on your own servers.

AI models: OpenAI, Claude, Gemini

The AI step is where the reading and understanding happens. The three providers we use most are:

• OpenAI (GPT models): a strong all-rounder with a very wide ecosystem, good for extraction, classification, and writing.
• Anthropic (Claude): strong at reading long documents and following detailed instructions carefully.
• Google (Gemini): works well with documents and images and fits naturally with Google Workspace. We used Gemini on Vergo to read contractor invoices and reports and compare bids.

Models improve every few months, so we do not pick one on reputation. We run the same real documents through each candidate and compare accuracy, speed, and cost. A well-designed workflow can also switch providers later without a rebuild.

Tools that do the work

An automation is only useful if it ends in a real action. These are the tools we connect most often:

• Google Calendar: book meetings and appointments automatically, and check availability before offering a time.
• Twilio: send SMS and messaging reminders and confirmations.
• SendGrid: send reliable transactional email such as confirmations, reports, and alerts.
• Stripe and PayPal: start workflows when a payment arrives, fails, or is refunded, and keep records in sync.
• Your own systems: CRM, accounting, spreadsheets, or a custom database connected through their APIs.

When custom code is better

No-code platforms are fast, but some jobs outgrow them. Custom code, which for us usually means Laravel and React, is the better choice when:

• the workflow handles a high volume of records and usage-based pricing gets expensive,
• the logic is complex, with many rules, roles, or approvals,
• people need a proper interface to review and act on the results, not just notifications,
• the data is sensitive and must stay inside your own systems.

Many of our projects mix both: n8n or Make for connecting outside apps, and a custom app for the core of the business.

How to choose

1. Start from the task, not the tool. Write down the trigger, the decision, and the action you need.
2. Check your existing apps. Make sure the platform connects to the tools you already use.
3. Estimate the volume. How many runs per month? That decides which pricing model is cheapest.
4. Test the AI step on real examples. Twenty real documents tell you more than any feature list.
5. Plan who maintains it. Automations need small fixes as your apps change, so decide who owns them.

Common questions