Demo
Every new Shopify order is checked against its Stripe payment, logged as a closed-won deal in Salesforce, added to the right Mailchimp list, and announced to the accounts team in Slack. Watch AnnE build it from one paragraph: she lays out the 12 steps, sets how the job starts, asks only what she can't know, and tries it on the store's orders before anything is written.
Fetch all new Shopify orders, verify the payment status in Stripe, look up or create the corresponding account and contact in Salesforce to log an closed-won opportunity, conditionally add the customer to a Mailchimp re-engagement campaign if they spent under $150 or a VIP list if they spent more, and instantly alert the accounts team in Slack with the Salesforce record links and invoice details
Shopify
Stripe
Salesforce
Mailchimp
SlackThe 12 steps, in 8 stages.
The job starts on a schedule and reads the Shopify orders that are new since the last run.
Orders from Shopify and payments from Stripe are joined, so every order carries its payment status.
A Keep only step lets through the orders whose payment Stripe confirms; the rest go no further.
Each order is shaped for the systems it goes to: the customer, the amount and the invoice details.
One Copy step sends every order to all the writes at once, so no system waits for another.
The account and the contact are looked up or created, and a closed-won opportunity is logged.
Under $150, the customer joins the Mailchimp re-engagement campaign; $150 or more, the VIP list.
Each order is posted to Slack with its Salesforce record links and invoice details.
Preview is a trial run of the whole job. Each step reads the real records of the system it is connected to, or its sample data where it is set to use samples, and keeps what it made for you to look over. Steps that write or send sit it out unless you turn them on.
A step still missing a setting is marked Not run, and the note above the canvas says which one and why; the steps after it wait. Here the Shopify read, the Stripe read and the join ran.
Rows in, rows out and the time each step took sit above it: 49 orders from Shopify in 820 ms, 6 charges from Stripe in 453 ms.
The panel under the canvas has a tab for each step that ran. Its rows read like a spreadsheet: search them, switch to the raw records, download them, or open the full page.
Previews and live runs alike are in Runs: the time, the rows sent across all steps, the failures, the slowest step and each step's numbers, with View logs, Rerun and Open in editor.
Shown in the job from the recording and in a larger one AnnE built: a month-end reconciliation across Shopify, Stripe, Square, Mailchimp, Google Sheets, Snowflake, QuickBooks, Slack and Gmail. Every part is a step you can open, read and change on the canvas.
Build steps
Each part of the request becomes a step: a read from a system, a join, a filter, an AI step, a total, a write, an approval. The steps are wired in the order the work happens, and every one opens on its own settings.
Formulas
Rules, totals and calculated fields are Excel-style formulas over the columns that reach the step. AnnE writes them from your request, and the platform checks each one against the real columns before the process runs.
Mapping
Shopify, Stripe and Square name the same thing differently. The join matches each source on its own order column, and a Map step shapes each row for the system it is written to.
Filtering
A Keep only step lets through the rows its formula allows, here the transactions created last month. The rest go no further, and the run reports how many.
Routing
A Route step sends each row down the path whose condition it meets: orders over $500 to the finance approval, everything else straight on. A Copy step sends the same rows to several targets at once.
Human approval
The approval step pauses the run before QuickBooks and notifies the approvers by email or Slack. They decide signed in, in the Worklist; the decision is kept with the rows, and only an approval is posted.
Preview
Preview plays the whole process on sample or connected data and shows the rows each step produced. You choose how the approval goes; steps that would write to QuickBooks, Slack or email sit the preview out unless you turn them on.
Schedules and webhooks
The order process starts on each paid Shopify order, and the reconciliation on the first of every month. A process can also start when its webhook is called, when a form is submitted or when someone starts it by hand.
Tables
Keep reference data, results or a running ledger in tables inside Ductile AI, read and written by any process.
Forms
Build a form and share its link; every submission starts a process with the answers as its first row.
From one paragraph she puts every step on the canvas, wired in the order the work happens, before she fills in any of them.
Asked for new orders, she sets the job to check every 5 minutes for anything new, and says so beside the step.
Which Salesforce connection to write with, and which orders go to Salesforce and to Slack when “every order” names no rule. She offers to read a few Stripe records, read-only, to fill in the column names.
A preview reads the connected records, skips the steps that write or send, and shows the rows each step produced, like a small spreadsheet.
Say it or type it in plain language. AnnE builds it, previews it with you, and runs it the same way every time.
Get started