Make: Visual Automation With Room to Grow

Make, formerly Integromat, runs automations on a visual canvas. How scenarios recover from errors, why they switch themselves off, and how credits are counted.

Make is a visual automation platform: workflows are drawn as a chain of connected modules rather than written as code. It began as Integromat, launched in Prague in 2012, and has been part of the process mining company Celonis since 2020. Its headquarters are still in Prague.

It connects to more than 3,000 apps, and an HTTP module reaches almost anything else with an API.

How scenarios work

A workflow in Make is a scenario. It starts from a trigger, either instantly when a webhook arrives or on a schedule, every 15 minutes by default. From there, routers split the flow into branches and filters decide what passes. Iterators and aggregators break lists apart and gather them back up, so one order’s line items can be handled one at a time.

Data stores act as small databases inside Make, which helps when a scenario needs to remember what it has already processed.

That memory matters for orders. Shopify’s developer docs note that the same webhook can arrive more than once, after a network timeout or a retry, and Make runs instant webhooks in parallel by default. A scenario that turns each Shopify order into a Xero invoice can start twice for one order. Checking a data store of order numbers before the Xero step stops the second invoice, and the Process data in order setting makes each run wait for the one before it.

Handling errors

Make gives each module its own error handlers: skip the item, retry, resume with a fallback value, or commit or roll back what has run so far. Failed runs can be kept as incomplete executions, so they can be fixed and finished later rather than lost.

When a scenario switches itself off

Every scenario has an errors before deactivation setting: the number of failed runs in a row Make accepts before it switches the scenario off. A scenario started by a webhook ignores that number. It’s deactivated at its first unhandled error.

By default, Make sends an email when an error stops a run and another when it deactivates a scenario, then a digest a few minutes later if more errors follow. Errors that an error handler catches, or that become incomplete executions, show up as warnings, which don’t count toward deactivation.

Leads and orders that arrive while a scenario is off wait in the webhook’s queue. How many it holds depends on the credits in the plan, and once it’s full, Make turns away anything new.

Credits

Since August 2025, Make has billed in credits, which replaced operations one for one. Each module run uses a credit, while triggers count once however many items they return. Some AI features use more, depending on what they process.

Keeping connections authorized

A connection is Make’s saved login to another app, and several scenarios can share one. Make lists expired API keys and access revoked after team or permission changes among the causes of connection errors, so one change of staff can stop every scenario that relies on that login. Whoever creates a connection gets an email the first time a teammate adds it to a scenario, and a connection can be replaced across several scenarios at once.

Shopify has a rule of its own. Make’s Shopify guide warns against setting up two connections for the same app and store, because the newer one overwrites the permissions of both.

Hosting and security

Each organization picks a data center in the EU or the US when it’s created. Make has completed SOC 2 Type II and SOC 3 audits, and its security program is ISO 27001 certified. It follows GDPR.

AI agents and assistants

Make launched AI Agents on paid plans in April 2025, and Make Grid, a live map of every scenario across an organization, went into open beta that June. Its MCP server lets AI assistants run Make tools, and Make became a built-in connector in Claude in March 2026. Maia, which builds scenarios from a plain description, reached public beta in August 2026.

An official Make plugin for ChatGPT followed in September 2026, and automations built through it keep running in Make’s cloud whether or not ChatGPT is open.

Where Make fits

Make reaches more than 3,000 apps, so it can sit in the middle of a lot of workflows. Shopify orders can become Xero invoices with a data store guarding against duplicates, new leads can be routed to the right person, and form entries can reach the CRM and a sheet at once.

Make’s error handling, with separate routes for different kinds of failure, and its pricing at volume both come up in the signs of outgrowing low-code tools. Once a scenario carries orders or invoices the business depends on, decide who owns it day to day before the first deactivation email lands.

Ready
to GLUE?

Got a bottleneck slowing you down? Get in touch to discuss your pain points and how we can streamline your workflow