Resources · Zapier integration

Databricks Zapier integrationAutomate Databricks with Zapier.

The Databricks Zapier integration gives you 3 actions and 10 searches, but no trigger. Other apps start the Zap; Databricks runs a query, a job, a pipeline or a Genie question and hands back the answer. Here is what each step does, what it costs and what breaks.

See the 13 operations

Verified Trustpilot reviews · AI, automation & growth agency

Overview

What the Databricks Zapier integration actually does

It lets everyday business tools ask Databricks for something, without anyone opening a notebook. A form, a chat message or a CRM update starts the Zap; Databricks runs SQL, kicks off a job or a pipeline, or answers a plain-language question through Genie, and the result flows back to the next step.

Three outcomes show up again and again. Numbers delivered where people work: Query SQL Warehouse runs a short query and passes the rows to a sheet, an email or a chat message. Data jobs launched from outside: Trigger Job and Trigger Pipeline start existing work in your workspace when something happens in another tool, so nobody has to log in to press run. Questions in plain words: Ask Genie a Question, then Get Genie Answer, let a non-technical colleague query a Genie space and get the reply back in their usual app.

Now the limits. Databricks has no trigger on Zapier: a Zap cannot start when a job finishes, a table changes or a pipeline fails. Most of the work happens in searches, and any search adds a step, so these Zaps quickly need a paid plan. Long queries and Genie answers do not come back at once either: they are built as two steps with a wait in between. If you need events coming out of Databricks, compare with the Databricks n8n integration and the Databricks Make integration.

On the plus side, Databricks is not a premium app on Zapier.

Vocabulary

Zapier vocabulary in one minute

Eight words you will see in the editor. A quick read now saves confusion later on this page.

Zap
The automation you build: one starting event, then one or more steps that run each time that event happens.
Trigger
The event that starts the Zap. With Databricks it always comes from another app, such as a new form entry.
Action
What the Zap does after it starts: here, running a statement, a job or a pipeline in Databricks.
Task
The unit Zapier bills. Each action that completes successfully counts as one task on your plan.
Polling vs instant
A polling trigger is checked on a timer. An instant trigger is pushed by the app as soon as the event happens.
Filter
A step that lets the Zap continue only when the data meets your condition, and stops it quietly otherwise.
Search step
A step that fetches something from an app, so later steps can use it. Most Databricks queries live here.
Connection
Your Databricks account linked to Zapier once, then reused by every Databricks step in every Zap.
Cost

What a Databricks Zap really costs

The trigger is free. Zapier bills each action that succeeds, one task each.

  • Free plan100 tasks a month
  • Steps per Zap on Free2 (trigger + action)
  • Databricks appNot premium
  • Databricks triggersNone, start from another app

A failed action is not counted against your plan. The free plan includes 100 tasks a month and unlimited Zaps, capped at two steps: one trigger, one action. That is enough for Trigger Job after a form entry. But a search, whether a SQL query or a Genie question, turns the Zap into a multi-step Zap, and so does a filter or a second action. Since nearly every useful Databricks pattern uses a search, plan on a paid plan. The interval grid depends on the trigger app, not on Databricks.

Connect

Connect Databricks to Zapier in three steps

  1. 01

    Add the connection in Zapier

    Go to Apps in your Zapier account and click + Add connection. Search for Databricks in the dialog and select it. Zapier publishes a separate article on administrator prerequisites, so check with your workspace admin before this step.

  2. 02

    Sign in to your workspace

    Click Add connection. A new tab opens on Databricks: sign in with an account that can reach the warehouses, jobs, pipelines and Genie spaces your Zaps will use. Anything you do not own needs read or write access granted to that account first.

  3. 03

    Grant access and reuse it

    Allow Zapier to use the account. The connection then appears on your Apps page, and every Databricks step can pick it. Zapier also links an article on administrator prerequisites for this connection: read it with your admin if the sign-in is refused.

First Zap

Your first Databricks Zap

GoalA sales manager fills a short form, Databricks runs the matching SQL query, and the figures land in the team chat.

  1. 01

    Choose the starting app

    Start a new Zap and pick the app where the request comes from, here a form tool. Databricks has no trigger, so the Zap always starts in another app.

  2. 02

    Test the trigger

    Run the test. Zapier brings back the three most recent submissions from the connected account, so you map the next steps on a real request.

  3. 03

    Add a filter

    Add a Filter step so the Zap only runs when the form carries a region. This makes it multi-step, which needs a paid plan.

  4. 04

    Query the warehouse

    Add Query SQL Warehouse, choose the warehouse and write the statement, with the region passed as a parameter rather than pasted into the text.

  5. 05

    Send, test, publish

    Add the chat step, map the query results into the message, test every step, then click Publish. Only requests sent after that are processed.

Actions

Databricks actions

Databricks gives you 3 actions. For each one: what it does for you, when to reach for it, and what to watch out for.

1

Create SQL Execution

Action

In Zapier“Submits a SQL statement to a Databricks SQL Warehouse without waiting for it to finish. Returns immediately with a Statement ID. Use this for large queries that would otherwise time out, then add a Delay step and use "Retrieve SQL Execution" to retrieve the data.”

Sends a SQL statement to a warehouse and moves on straight away, keeping only a Statement ID. The query keeps running in Databricks while the Zap waits.

When to use it
heavy queries that would time out if Zapier waited for the result; fetch the rows later with Retrieve SQL Execution.
Watch out
put a Delay step between the two, or the result will not be ready yet.
2

Trigger Job

Action

In Zapier“Triggers a job to run.”

Starts a run of an existing job in your workspace. You pick the job from a list or paste its ID; the run itself happens in Databricks.

When to use it
refresh a report or a model when a new file lands, or when a manager asks for it from a form.
Watch out
no Databricks trigger reports when the run ends, so the Zap cannot react to its result.
3

Trigger Pipeline

Action

In Zapier“Starts a new update for a Delta Live Table pipeline.”

Launches a new update of a Delta Live Table pipeline. Two optional switches change its behavior: a full refresh that rebuilds every table, and a validation-only mode that checks the request without running it.

When to use it
update a pipeline when upstream data is ready, signaled by another tool.
Watch out
Full Refresh resets all tables first, so leave it off unless you mean it.
Searches

Databricks searches

10 search steps look for data that already exists in Databricks, so a later step can use it.

Ten of the thirteen Databricks steps are searches: they sit at the bottom of the action list, under the SEARCH heading, and return data to the next steps. Three settings decide what happens when a search comes back empty or with several results.

1

Find Tables

Search

In Zapier“Search for tables in Unity Catalog by name, description, or column names using keyword matching.”

Looks through a Unity Catalog for tables whose name, description or columns match your keywords, then returns what it found.

When to use it
help an analyst locate the right table from a chat request before writing any SQL.
Watch out
you must pick the catalog first, and matching is by keyword, so vague words bring back noise.
2

Ask Genie Follow-Up Question

Search

In Zapier“Ask a follow-up question in an existing Genie conversation, keeping context from previous questions. Add a Delay step, then use "Get Genie Answer" to retrieve the response.”

Adds a new question to a Genie conversation that is already open, so Genie keeps the context of what was asked before.

When to use it
drill into a first answer, for example by asking for the same figure split by region.
Watch out
map the Conversation ID from the first question step, use the same Genie space, and fetch the reply later with Get Genie Answer.
3

Get Genie Answer

Search

In Zapier“Retrieve the answer from a previously submitted Genie question. Use after "Ask Genie a Question" (or "Ask Genie Follow-up Question") with a Delay step (60–120s) in between.”

Collects the reply to a question already sent to Genie, using the space, the conversation and the message identifiers from the earlier step.

When to use it
as the second half of any Genie pair, right before the step that shares the answer.
Watch out
the app's own guidance pairs it with a wait of 60 to 120 seconds after the question, otherwise the answer may not exist yet.
4

Get Genie Space Details

Search

In Zapier“Look up a Genie space to see its configuration, available tables, and warehouse info. Useful for validating a space before querying.”

Reads the setup of a Genie space: how it is configured, which tables it can use and which warehouse sits behind it.

When to use it
check a space is ready before a Zap starts sending it questions, or log its settings for an audit.
Watch out
select the space from the list or paste its ID or full link; the space is the only required setting.
5

Query Online Table

Search

In Zapier“Query Online table.”

Reads from an online table through its endpoint and returns what it finds to the next step.

When to use it
fetch the latest score or attribute for one customer when a ticket or a lead arrives in another tool.
Watch out
the only required setting is the endpoint name, so make sure your team shares the exact spelling.
6

Query Serving Endpoint

Search

In Zapier“Queries Databricks serving endpoint based on its name.”

Sends a request to a model served in Databricks, picked by its endpoint name, and returns the model's output to the Zap.

When to use it
score an incoming lead or classify a support message with a model your data team already deployed.
Watch out
an optional Stream switch controls streaming for language models; leave it alone unless your team asks.
7

Query SQL Warehouse

Search

In Zapier“Runs a query on a Databricks SQL Warehouse.”

Runs a SQL statement on a warehouse and waits for the rows, then passes them to the next step. The simplest way to bring Databricks numbers into another app.

When to use it
short, fast queries, like yesterday's orders for one region.
Watch out
pass values as named parameters; for long queries, switch to Create SQL Execution.
8

Query Vector Search Index

Search

In Zapier“Runs a query against Vector Search Index.”

Searches a vector index with a piece of text and returns the closest matching records, with the columns you choose.

When to use it
find similar past tickets or documents when a new request comes in.
Watch out
five settings are required, including endpoint, index, columns and query type (ANN by default, or HYBRID), so set it up with whoever built the index.
9

Retrieve SQL Execution

Search

In Zapier“Retrieve the results of a previously submitted SQL execution. Use after "Create SQL Execution" with a Delay step (60 - 120 seconds) in between.”

Picks up the results of a statement started earlier with Create SQL Execution, from its Statement ID.

When to use it
the second half of a long query, after a Delay step of 60 to 120 seconds.
Watch out
map the Statement ID from the earlier step rather than typing it, and give large queries enough time before this step runs.
10

Ask Genie a Question

Search

In Zapier“Ask a natural language question to a Databricks Genie space. Returns immediately with a conversation ID. Add a Delay step (60–120s), then use "Get Genie Answer" to retrieve the response.”

Opens a conversation with a Genie space by asking a question in everyday language. Genie writes and runs the SQL itself, and the step returns a conversation ID right away.

When to use it
let a sales or ops lead ask about the data from a form or a chat.
Watch out
the answer comes later, through Get Genie Answer.

The three settings every search exposes

  1. Successful if no search results are found?
    Off by default: an empty result stops the Zap and skips the steps that depended on it. Switch it on to let the Zap carry on with nothing.
  2. Create X if it doesn't exist yet?
    This box turns a search into find-or-create. No Databricks search offers it, so nothing gets created when a lookup fails.
  3. If multiple search results are found?
    Three options: keep the first result (the default), stop the Zap, or pass all results on together.
When it breaks

When the Databricks Zap breaks

Zapier links two help articles to Databricks: a getting-started guide and one on administrator prerequisites for the connection. No Databricks-specific rate limit is listed.

Most failures come from timing and permissions, plus the general way Zapier works. The six cases below are the ones to plan for. If you are already stuck, the Zapier troubleshooting guide goes through the fixes.

  • The long query times out

    Query SQL Warehouse waits for the rows. On a heavy query, switch to Create SQL Execution, add a Delay step, then Retrieve SQL Execution.

  • Genie has not answered yet

    Get Genie Answer runs too soon after the question and finds nothing. The app's own guidance is a wait of 60 to 120 seconds in between.

  • Admin prerequisites are missing

    Zapier has a dedicated article on what a Databricks administrator must set up before the connection works. Read it with your admin before blaming the Zap.

  • A full refresh wipes the tables

    Trigger Pipeline with Full Refresh on resets every table before it runs. Leave the switch off, or test with Validate Only first.

  • Past requests are never processed

    A Zap only handles what happens after you publish it. Earlier form entries or messages stay untouched unless you replay them yourself.

  • A bulk import gets held back

    If the starting app suddenly returns a huge batch after an import, Zapier's flood protection can pause those runs before they reach Databricks.
Need help

Need help automating Databricks with Zapier?

A person reads every message.

FAQ

Databricks and Zapier: common questions

01Is the Databricks Zapier integration free?
Partly. Databricks is not a premium app on Zapier, so the free plan can connect it, with 100 tasks a month. The free plan only allows two-step Zaps, though: one trigger and one action. A form that starts Trigger Job fits. But most Databricks patterns rely on a search, such as Query SQL Warehouse or a Genie question, and a search makes the Zap multi-step. Those Zaps need a paid plan. The same goes for a filter or a second action.
02How many tasks does a Databricks Zap use?
One task for each action that completes successfully, and nothing for the trigger. A Zap that triggers a job uses one task per run. A Genie pattern with a question and then an answer adds a step for each, and a message that shares the result counts as well. Waiting for the starting app to report something costs nothing, however often Zapier checks. A step that fails is not counted. Keep Zaps short: every extra action that succeeds is billed on each run.
03Does the Databricks integration work in real time?
Not on its own. Databricks has no trigger on Zapier, so the timing depends on the app that starts the Zap. With an instant trigger there, the Zap starts at once on any plan. With a polling trigger, Zapier checks every 15 minutes on Free, 2 minutes on Professional and 1 minute on Team and Enterprise. Then there is Databricks itself: long queries and Genie answers are split in two steps with a wait of 60 to 120 seconds between them.
04What does it take to connect Databricks to Zapier?
A Databricks account that can reach the warehouses, jobs, pipelines and Genie spaces you want to use, with read or write access to anything you do not own. Zapier also publishes an article on administrator prerequisites, so involve your workspace admin early. The connection itself starts on the Apps page of Zapier with + Add connection, followed by a sign-in in a new tab. Have your Warehouse ID at hand for the SQL steps.
05Why does Get Genie Answer return nothing?
Usually because it runs too early. Ask Genie a Question returns a conversation ID immediately, but Genie still has to write and run the SQL. If the next step asks for the answer right away, there is nothing to collect yet. The app guidance is to place a Delay step of 60 to 120 seconds between the question and Get Genie Answer, and to map the Conversation ID and Message ID from the question step. The same pattern applies to SQL executions.
06What if the Databricks operation you need does not exist?
Then this Zapier app cannot do it today. There is no trigger at all, so a Zap cannot react to a finished job, a failed pipeline or a new row. There is no action to create tables or manage permissions either. Sometimes Query SQL Warehouse covers the gap, since it runs the SQL statement you write. If you need events coming out of Databricks, n8n and Make each have their own Databricks integration: compare their operation lists with this one.
07Zapier, Make or n8n for Databricks?
Pick by criteria, not by ranking. On Zapier, Databricks offers the 3 actions and 10 searches on this page and no trigger, so a Zap always starts in another app, and most useful patterns need a paid plan because a search adds a step. Make and n8n each have their own Databricks integration, with their own operations and their own way of billing a run. Compare three things: whether the operation you need exists in each tool, how a run with several steps is billed, and who on your team will maintain it.
Same platform

Other docs you might need on this stack.

Browse all integrations