Resources · Zapier integration

Function Zapier integrationAutomate Function with Zapier.

The Function Zapier integration gives you 1 action, 0 triggers and 0 searches. One move, Predict, runs a cloud predictor on data another app hands over. Function never starts a Zap on its own, so something else always goes first.

Verified Trustpilot reviews · AI, automation & growth agency

Overview

What the Function Zapier integration actually does

It lets a Zap (an automation on Zapier that starts on one event, then runs one or more steps) send data to a Function predictor and get a prediction back. Function is the AI step in the middle. It cannot be the starting point, because Zapier lists no trigger for it, only the Predict action.

Think of it as a scoring desk you can bolt onto any app Zapier already knows. Three results people usually chase with it:

A scored form answer. A survey or signup arrives, Predict runs the predictor you picked by its tag, and the result lands next to the answer in a spreadsheet.

A labeled record. A new deal or ticket appears in another tool, Predict reads the text you mapped into it, and a later step writes the label back. Same action, different input.

A routed message. The prediction goes into a chat or email step, so the right person sees it without opening Function at all.

Now the limits. Zapier does nothing on the Function side except call a predictor. It cannot create or train one, it cannot list your predictors, and it only reaches cloud predictors. A predictor that lives anywhere else stays out of reach. There is no search either, so a Zap cannot look up an earlier prediction.

Comparing tools before you commit? The Make vs Zapier comparison is a good second read.

Vocabulary

Zapier words, in one minute

Seven terms come back on this page. Each one is explained once, here, then used as is.

Zap
The automation you build: one event that starts it, then the steps that run every time that event happens again.
Trigger
The event in an app that starts the Zap. Function has none, so a form, a CRM or a sheet plays that role.
Action
What the Zap does after it starts: the step that writes, sends or, here, asks a predictor for an answer.
Task
The unit Zapier bills. One action that completes successfully counts as one task on your monthly allowance.
Polling vs instant trigger
A polling trigger is checked by Zapier on a timer. An instant trigger is reported by the app itself the moment it happens.
Filter
A step that lets the Zap continue only when the incoming data meets your condition, and stops it quietly otherwise.
Connection
Your Function account, linked to Zapier once and then reused by every step that needs to call one of its predictors.
Cost

What a Function Zap costs in tasks

Function is not a premium app, so it works on the free plan. The bill is counted in tasks, not in predictions.

  • Premium appNo, free plan works
  • Free plan100 tasks a month, two-step Zaps only
  • Trigger checksFree, never billed
  • Failed PredictNot counted

The trigger costs nothing, however often Zapier checks it. Each Predict that succeeds counts as one task, and a failed one is not counted. The free plan gives 100 tasks a month but stops at two steps: one trigger, one action. Add a filter or a second action to store the result, and the Zap becomes multi-step, which needs a paid plan. In practice, budget one task per record for Predict, plus one more for every step that stores or sends the result.

Connect

Connect Function to Zapier in three steps

  1. 01

    Open the Apps page

    In your Zapier account, go to Apps and click + Add connection. You need a Function account first, plus access to any predictor you do not own yourself but want the Zap to call.

  2. 02

    Pick Function in the dialog

    Type Function in the search box of the dialog, select it, then click Add connection. A new tab opens on the Function side so you can sign in with the account that holds your predictors.

  3. 03

    Grant access

    Sign in and allow Zapier to use the account. The connection now shows on the Apps page, and every Predict step you add later reuses it without asking again.

First Zap

Your first Function Zap, step by step

GoalScore every new Typeform answer with a Function predictor and keep the result in Google Sheets.

  1. 01

    Create the Zap

    Start a new Zap and pick the app that starts it. Here a Typeform survey plays that part, since Function cannot open a Zap by itself.

  2. 02

    Set and test the trigger

    Choose the form, then test. Zapier brings back the three most recent answers, so you map fields from a real one rather than a blank example.

  3. 03

    Add Predict

    Add an action, pick Function, then Predict. Paste the tag of your cloud predictor, then fill the other fields the step shows with values from the Typeform answer.

  4. 04

    Store the result

    Add a Google Sheets row and map in what Predict returned. This third step makes the Zap multi-step, so it needs a paid plan.

  5. 05

    Test, then publish

    Run each step once and read what came back, starting with Predict. When the new row looks right in the sheet, click Publish to switch the Zap on.

Actions

The Function action in Zapier

Function gives you 1 action. For each one: what it does for you, when to reach for it, and what to watch out for.

1

Predict

Action

In Zapier“Make an AI prediction with a predictor.”

Hands the data you mapped to one of your Function predictors and brings its prediction back into the Zap, ready for the next step to store or send.

When to use it
a new lead, answer or ticket needs a score or a label before anyone looks at it.
Watch out
the tag field is required and only cloud predictors are reachable, so a predictor running elsewhere will not show up.
When it breaks

When a Function Zap breaks

Zapier attaches no help article to Function and no app-specific limit is documented. What breaks comes from how Zaps work in general.

Most surprises show up on the first day, not the hundredth. These four cover them, and the Zapier troubleshooting page picks up from there.

  • Old records are never scored

    Predict only ever sees answers submitted after the Zap is published. Answers that already existed stay unscored, and moving that backlog is a separate job.
  • The Zap keeps setting itself off

    If the trigger watches a sheet for new rows and the last step writes the prediction into a new row of that same sheet, the Zap restarts on its own result and burns tasks.
  • A bulk import gets held back

    When the trigger app suddenly returns a huge batch, after an import or a migration, Zapier's flood protection can hold those items back instead of running Predict on each.
  • No double predictions

    On a polling trigger, Zapier remembers what it has already seen, so the same record does not start the Zap twice and you do not pay twice for one prediction.
Need help

Need help automating Function with Zapier?

A person reads every message.

FAQ

Function Zapier integration: common questions

01Is the Function Zapier integration free?
Yes, as long as the Zap stays at two steps. Function is not a premium app, so the free plan accepts it, with 100 tasks a month. The catch is the step limit: the free plan builds a trigger plus one action, nothing more. A Zap that starts in a form and then runs Predict fits. A Zap that also writes the prediction somewhere, or filters the input first, is multi-step and needs a paid plan. Most useful Function Zaps end up there, because a prediction nobody stores is hard to use.
02How many tasks does a Function prediction use?
One task per Predict that succeeds. The trigger never costs anything, and neither do the checks Zapier runs to find new data. If Predict fails, the attempt is not counted against the plan. Every other action that succeeds counts on its own, so a Zap with Predict plus a spreadsheet step uses two tasks per run. Plan the monthly allowance on that basis: the number of records coming in, times the number of actions that complete for each one. A filter placed before Predict keeps useless records from eating tasks.
03Does Function run in real time in Zapier?
It depends on the trigger, because Function has none of its own. Predict runs as soon as the Zap reaches it, so the delay comes from the first app. If that app reports its events instantly, the whole chain moves straight away, whatever the plan. If its trigger is a polling one, Zapier checks on a timer: every 15 minutes on Free, every 2 minutes on Professional, every minute on Team and Enterprise. Picking an instant trigger is the simplest way to get predictions fast on a free plan.
04What is needed to connect a Function account?
A Function account, plus access to any predictor you want to call but do not own yourself. The connection is made once from the Apps page with + Add connection, then you sign in to Function in the tab that opens and allow access. After that, every Predict step reuses it. Before connecting, it helps to know the tag of the predictor you plan to use, because the action will not run without it, and to confirm that this predictor is deployed in the cloud.
05Why does a Function predictor fail in Zapier?
Most likely because it is not a cloud predictor. The Predict action only reaches predictors deployed in the cloud, so a model running anywhere else cannot be called from a Zap. The second usual cause is the tag: the field is required and has to match the predictor exactly. A typo or an outdated tag stops the step. Test Predict on its own in the editor first, read what comes back, and only then add the steps that store or send the result.
06What if the Function operation needed is missing?
Zapier offers one operation for Function, Predict, and nothing else. There is no trigger, so Function cannot start a Zap, and no search, so a Zap cannot look up an earlier prediction. Managing predictors, training them or listing them is not possible from Zapier either. The workaround is to keep Function in the middle of the Zap: another app starts it, Predict does the AI work, and a third app keeps the result. Anything beyond that has to happen inside Function itself.
07Zapier, Make or n8n for Function?
Judge it on criteria rather than brand. On Zapier, Function brings a single action, Predict, limited to cloud predictors, with no trigger and no search. It works on the free plan for a two-step Zap, and the cost is one task per successful prediction. So the questions are: does the other tool offer a Function connector at all, does it reach the predictors you use, and how does it bill a run? If your flow needs more than two steps, compare the paid tiers on that basis.