AgentsStudioTemplatesPricingDocs
Sign inGet Started

Give your AI agent an automation engine.

Build

  • MCP server
  • kovca CLI
  • Kovca's agent
  • Studio
  • Templates

Run

  • Apps and events
  • Decision and Generative AI
  • Workflows API
  • Run history
  • Pricing

People in the loop

  • Approvals and waits
  • Tasks
  • Coding agents
  • Docs

Company

  • About
  • Security
  • Contact
  • Privacy Policy
  • Terms of Service
© Kovca. kovca.io
GuidesOperations

Look up where-is-my-order emails in Shopify and label them for a person

Build a Kovca workflow that starts from a new Gmail message, looks an order number up in Shopify, and labels the message for a person. Reply is an optional draft that names only that number.

Support Triage28 min readcURL
InputSupport TriageExtractHas order num…Find ordersCountOne order?Reply textReplyNeeds personResult
TemplateGmail Shopify order statusStart from a new Gmail message, look up an order number in Shopify when the message is filed as other and states one, and label it for a person. Reply is an optional draft that names only that order number.
11 nodesUp to 1 evaluation per run
Use this template

A where-is-my-order email lands in a shared support inbox. This guide builds a workflow that starts when a new Gmail message arrives, classifies the subject and the body, copies an order number when the email states one, looks that number up with Shopify Find orders, and applies a label so a person can answer. That lookup plus the label is what the Studio test shows. You'll start from a template and test it with simulated actions.

The template already uses the Gmail New email trigger, Shopify Find orders, Gmail Reply, and Gmail Add label. It does not store a Gmail or Shopify account. You choose the accounts when you activate. Until then the draft saves, and Studio tests simulate the Gmail and Shopify steps.

Reply is an optional draft. The recorded test does not send it, and this guide does not show a sent reply. The draft names only the order number copied from the email. It does not quote fulfillment or tracking. Enable it only after a lookup on a store you control has shown you a real Find orders result, and only when the sentence is one you want a customer to read.

Support Triage does not have an order-status category. It returns one of billing, technical, account, product, or other, plus priority and requires_human_review. In the node's own wording, other is something else, such as shipping, partnerships, job inquiries, or spam. This template treats other plus an order number the email states as the lookup path. A refund question is billing. A bug report is technical. Partnerships, job inquiries, and spam can also be other, so an order number in that mail still takes the lookup path.

Find orders returns ok and result. result is everything the action returned. The catalog does not name a list of orders, a count, a fulfillment status, a tracking number, or a tracking URL inside result. The sample result is {}. Count counts result only when that value is a list. {} is not a list, so the count is null, One order? is false, and Reply does not run. The message gets the Needs a person label. That is what a Studio test shows with Send real requests off.

Shopify's triggers in the Studio are product, stock, collection, and customer-segment events. There is no order trigger, which is why this workflow starts from email. The Shopify app's read_orders access covers the last 60 days of orders. An older order is outside that window.

Reply sends for real once you choose a Gmail account and turn Send real requests on, or once the workflow is published and activated. The step's own line says it is sent once: Kovca never retries it automatically. Add label is safe to retry. Test with simulated actions first. The reply in this template does not run on the sample result, which is the label-only path. Leave the Gmail account off Reply until a lookup on a store you control has shown you the shape of result, and until the reply text is a sentence you want a customer to read.

Label a shared Gmail inbox by topic and urgency applies a topic label from the same Support Triage categories and does not look up an order. Catch Shopify variants going out of stock and start a restock follow-up starts from stock, not from an email.

If the inbox only gets a few of these questions a day, a saved Gmail reply that points people at the store's order status page, or a helpdesk that already shows the order next to the email, is less machinery than this.

What you'll build

Send real requests is off. The run returned 200 and used 6 credits, with 558 Generative AI tokens. category is other, order_number is #1042, status_email is true, order_count is null, lookup_result is an empty object, one_order is false, reply is skipped, and needs_person is marked sample.

StepNodeWhat it does
InputGmail: New email (gmail/new_email)Starts the workflow. The event fields are from, body_text, received_at, labels, message_id, thread_id, to, and subject.
Support TriageDecision AI (builtin:support/triage@1)Classifies the subject and the body. Returns category, priority, and requires_human_review.
ExtractGenerative AI (ai:extract@1)Copies order_number when the email states one. Leaves it out otherwise.
Has order number?ConditionTrue when category is other and order_number is present.
Find ordersShopify Find orders (action:shopify/find_orders@1)On the true branch. Filters by the order name, with status any. Safe to retry.
CountListCounts result when it is a list. Otherwise the count is null.
One order?ConditionTrue when that count equals 1.
Reply textTextThe reply sentence, on the true branch. It names the order number and does not quote fulfillment or tracking.
ReplyGmail Reply (action:gmail/reply@1)Sends that text on the thread. Sent once.
Needs personGmail Add label (action:gmail/add_label@1)On either false branch. Applies one stand-in label id. Safe to retry.
OutputOutputThe card on the canvas says Result. Returns the classification, the order number, the count, and whether the reply or the label ran.

Both false branches go to Needs person, then Result. A message that is not other with an order number never calls Shopify. A lookup whose result is not a list of one item does not send a reply.

Watch the test

Watch the Studio test (MP4). It is a recording of the production Studio at kovca.io, in light mode, 23 seconds. The frame is the Test panel, with no browser chrome. It starts on a clear panel, with no earlier result on screen. The input includes message_id. Send real requests stays off. The clip runs the test. A pause after the result appears, before the panel is scrolled to the whole output, is cut. Nothing that remains is sped up. The end of the clip holds on the full output, including needs_person true and the closing brace. No Gmail or Shopify account is connected. Find orders and Needs person return their samples and are marked sample. That is not an order loaded from a store, and it is not a label applied in Gmail. Reply is skipped.

The full output object, including needs_person true and the closing brace. category other, order_number #1042, status_email true, order_count null, lookup_result empty object, one_order false, reply_text null, replied null.

The recording follows these steps:

  1. The Test tab is open on a clear panel. The input is the order question: subject "Where is my order #1042", body "Where is my order #1042? Has it shipped? I need the tracking number.", from alex@example.com, to support@example.com, with message_id 18f0a1b2c3d4e5d1 and the same value as thread_id. Send real requests is off. The switch's own text says action steps return their sample result and nothing is sent.
  2. Run test. The status is 200. The panel shows 6 credits used (1 Decision AI evaluation) and 558 Generative AI tokens.
  3. The category is other, the priority is low, and requires_human_review is false. order_number is #1042. status_email is true.
  4. Find orders succeeded and is marked sample. lookup_result is {}. order_count is null. one_order is false. Reply text and Reply are skipped. Needs person succeeded and is marked sample. needs_person is true. reply_text and replied are null.

That is one live Studio test of that email. Extract returned #1042. Support Triage returned other. Find orders returned the sample {}, which is not a list, so Count did not return 1 and the reply did not run. What Support Triage and Extract return on another message depends on that message.

The JSON later in this guide is the template's unit test for the same input when the test does not read the message. The category is the first Support Triage category, billing, and Extract returns no order number, so Find orders is skipped and Needs person is the only action that runs.

Build it in the Studio

1Start from the template

Click Use this template above, or open Gmail Shopify order status in Workflows.

The draft opens on the canvas. The components library is on the left. Triggers, AI, Action in an app, Data transformation, Flow, Core, and Human review are the groups. This workflow uses Gmail and Shopify, so both are under Triggers and under Action in an app once the draft uses them. Apps you have not connected, and that this workflow does not use, stay behind Connect apps. The pictures below are the production Studio, in light mode, with no Gmail or Shopify account connected. The template's step titles are the ones in the table above. The last card on the canvas is Result.

Triggers, Gmail. New email fires when a new message is received in Gmail.

Product created, product updated, variant out of stock, variant back in stock, inventory level updated, collections, and customer segments. No order trigger.

Get order, Find orders, Update order, Get customer, Get a customer's orders, and Get product.

Add label, Remove label, Mark as read, Archive, Reply, and Send email.

2The workflow starts from a new email

Select the input step. It is Gmail: New email, not an API request. The account line asks you to connect Gmail. Leave it empty while you test.

Connect Gmail is still to do. Only emails matching is to:support@example.com -from:me. In label asks you to connect Gmail to choose a label.

SettingTemplate value
Only emails matching (query)to:support@example.com -from:me
In label (label_ids)unset

to:support@example.com -from:me is a stand-in. Replace it with the search you actually want. The field's description is "Advanced Gmail search using the same syntax as Gmail's search box." to: limits the search to mail for that address. -from:me leaves out mail from the connected account. A live message that does not match that search does not start a run. The Test tab and an API call send the event you provide, so they still run when the search would not have matched.

In label filters by a single label id. Until an account is connected, the field says to connect Gmail to choose a label. The template leaves it unset.

The trigger description says it fires when a new message is received in Gmail. The same message does not start two runs: the event id is message_id. A later message is a new event and runs again.

Event fields later steps read:

FieldUsed for
input.subjectSupport Triage's subject, and Extract
input.body_textSupport Triage's message, and Extract
input.fromExtract, and Reply's To
input.message_idNeeds person
input.thread_idReply

The event also includes received_at, labels, and to. labels is a list of label ids already on the message. The trigger's sample uses INBOX and UNREAD. This workflow does not read that list. from is a string, the sender as Gmail reports it. Reply's To field asks for an address in the form user@domain.com. The examples in this template use a bare address. The trigger's own sample sender includes a display name. If the live from is not a bare address, change the To mapping before you activate. A manual run or an API call can omit fields the steps don't read.

subject on Support Triage accepts at most 200 characters. The event's subject can be up to 8,000. A longer subject fails this step. message is required, must not be empty, and accepts at most 8,000 characters, which is also the limit on body_text. An empty body fails Support Triage.

The body is what the node reads, along with the subject when you map it. Support Triage's question is about message. This template maps body_text to message and subject to subject. It does not read attachments.

3Classify the email

Select Support Triage. subject is input.subject. message is input.body_text.

subject is input.subject. message is input.body_text.

The node returns:

  • category: billing, technical, account, product, or other.
  • priority: low, medium, or high.
  • requires_human_review: true or false.

What each category means, in the node's own wording:

CategoryThe request is mainly about
billingCharges, invoices, payments, refunds, subscriptions, or pricing.
technicalErrors, bugs, outages, crashes, slow performance, or integration problems.
accountSigning in, passwords, account access, profile details, or account settings.
productHow a feature works, how to use the product, or a feature request.
otherSomething else, such as shipping, partnerships, job inquiries, or spam.

Shipping sits in other. So do partnerships, job inquiries, and spam. A refund sits in billing. Support Triage may file a real order-status question under another category, for example product. The lookup never runs in that case. This template does not add a separate rule for priority or for requires_human_review. A high message that is other and states an order number still takes the lookup path. Add a rule if those should go straight to Needs person.

4Copy the order number

Select Extract. It copies one field from the subject, the sender, and the body.

Instructions say to copy an order name or number the email states, such as #1042, and to leave order_number out otherwise. The field is text and is not required.

Extract the order number from this email. Copy only an order name or number the email states, such as #1042. Leave order_number out when the email does not state one. Do not guess, and do not invent a number.

Subject: {{input.subject}} From: {{input.from}} Message: {{input.body_text}}

FieldWhat it copies
order_numberThe order name or number the email states, such as #1042. Not required.

The field is left out when the email does not state a number. A missing order_number makes Has order number? false, and Find orders does not run. On the recorded Studio test, this step returned #1042.

5Decide whether to look the order up

Select Has order number?. A condition continues on true when every rule matches. This one has two rules. The name is the check this step makes. It is not a separate order-status category.

Continue on true when all rules match. category equals other. order_number is present. The title is Has order number?.

RuleValue
Continue on true when all rules matchnodes.triage.output.category equals other
nodes.extract.output.order_number is present
CaseWhenGoes to
trueother, and an order number was copiedFind orders
falseany other category, or no order numberNeeds person

Support Triage may file a real order-status question under another category, for example product. Has order number? is then false, and the lookup never runs.

The step's id is status. The output field status_email is that step's result.

6Look up the order

Find orders is Shopify's Find orders action, on the true branch. No account is chosen.

Connect Shopify is still to do. If it fails is Continue (output.ok is false). Order name is nodes.extract.output.order_number. Status is the fixed value any. Payment status, fulfillment status, created after, created before, and limit are not set.

Find orders fieldValue
Order name (name)nodes.extract.output.order_number
Status (status)any
Payment status (financial_status)unset
Fulfillment status (fulfillment_status)unset
Created after (created_at_min)unset
Created before (created_at_max)unset
Limit (limit)unset

The order name field's description is "Filter by order name (e.g., #1001)." Extract copies the number as the email states it, including a # when the email has one. Status any is one of the values the field's description lists (open, closed, cancelled, any). The template sets it so the lookup is not limited to a single status word. The fulfillment status field here is a filter on the query. It is not a value the action returns as its own output field.

If it fails is Continue (output.ok is false). A failed lookup still reaches Count. The count is null, One order? is false, and Needs person runs. The step is safe to retry: Kovca retries it if the app is briefly unavailable. In a Studio test, with Send real requests off, the step returns its sample and is marked sample. The sample is ok: true and result: {}. That is not an order from your store.

read_orders covers the last 60 days. Find orders does not add a setting that extends that window.

7Count the result, and reply only for a list of one

Count reads nodes.find.output.result and counts it when that value is a list. Any other shape, including {}, comes back null.

One order? is true when that count equals 1.

Continue on true when all rules match. nodes.count.output.result equals 1.

CaseWhenGoes to
trueresult is a list of one itemReply text, then Reply
falsenull, 0, or any other countNeeds person

The catalog does not say result is a list of orders. On the recorded test it was {}, the count was null, and this step was false. If a live lookup on a store you control keeps the orders under another key, point Count at that list after you have seen it. Until then this branch does not send mail.

Reply text runs only on the true branch.

Thanks for asking about order {{nodes.extract.output.order_number}}. A person will follow up with fulfillment and tracking.

Thanks for asking about order {{nodes.extract.output.order_number}}. A person will follow up with fulfillment and tracking.

That sentence uses the order number copied from the email. It does not read result, because the catalog does not name fulfillment status, a tracking number, or a tracking URL there. Replace the sentence after a real lookup shows you those fields, and only with paths you have seen. Until then, leave the Gmail account off Reply.

Reply sends that text.

Connect Gmail is still to do. If it fails is Stop the run. Thread ID is input.thread_id. To is input.from. Reply text is nodes.reply_text.output.text. Cc is not set.

Reply fieldValue
Thread ID (thread_id)input.thread_id
To (recipient_email)input.from
Reply text (message_body)nodes.reply_text.output.text
Cc (cc)unset

The action's description says it sends a reply to a Gmail thread using its original subject and quoting the last returned message. The step's own line says it is sent once: Kovca never retries it automatically. If it fails is Stop the run. A failed send stops the run, and Needs person is not on this branch, so a failed reply does not also apply the label. In a Studio test the step does not run when One order? is false. On the recorded test it was skipped.

8Label everything else for a person

Needs person is Gmail's Add label action. Both false branches connect to it.

The title is Needs person. Connect Gmail is still to do. If it fails is Continue (output.ok is false). Message ID is input.message_id. Label IDs is the fixed list Label_needs_person.

JSON
["Label_needs_person"]
Add label fieldValue
Message ID (message_id)input.message_id
Label IDs (add_label_ids)["Label_needs_person"]

Label_needs_person is a stand-in. It is not a label id from a mailbox. Add label's own description of the field is "Label IDs to add (IDs, not display names)." Replace it with an id from the mailbox once Gmail is connected. The display name you see in Gmail is not what this step sends.

The step is safe to retry: Kovca retries it if the app is briefly unavailable, and repeating it has no further effect. If it fails is Continue (output.ok is false). In a Studio test it returns its sample, ok: true, and is marked sample. That is not a label applied in your mailbox.

Result returns the fields the rest of this guide names.

The title is Result. category, priority, and requires_human_review from Support Triage. order_number from Extract. status_email from Has order number?. order_count from Count. one_order from One order?. lookup_result from Find orders. reply_text, replied, and needs_person from the steps that send or label.

9Test it

Open the Test tab. Where is order 1042 and Refund request are the template's examples. Sample new email is the trigger's own sample, which is an invitation, not one of these messages. Leave Send real requests off.

On the recorded run of the order question, Support Triage returned other and Extract returned #1042, so Find orders ran and returned the sample {}. Reply was skipped. Needs person was marked sample. A message Extract does not pull a number from, and a message Support Triage does not file as other, skip Find orders and only label.

Before you activate, run messages you have actually seen, including ones that should be answered by a person. When you have a store connected, turn on Send real requests once for Find orders only, with Reply still lacking an account, and read lookup_result. If that value is a list, Count can see its length. If the orders sit under a key inside the object, point Count at that list. Do not guess the key. Do not turn on Reply until the text quotes only fields you have seen.

Publish, call it, and activate it

The same workflow can be called over the API. The body uses the trigger's field names. Create an API key under API keys and keep it on your backend.

cURL
curl https://kovca.io/api/v1/workflows/wf_.../execute \
  -H "Authorization: Bearer $KOVCA_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: 18f0a1b2c3d4e5d1" \
  -d '{
    "input": {
      "from": "alex@example.com",
      "to": "support@example.com",
      "subject": "Where is my order #1042",
      "body_text": "Where is my order #1042? Has it shipped? I need the tracking number.",
      "message_id": "18f0a1b2c3d4e5d1",
      "thread_id": "18f0a1b2c3d4e5d1",
      "received_at": "2026-11-02T14:03:00Z",
      "labels": ["INBOX", "UNREAD"]
    }
  }'

That curl calls a published workflow. It runs Support Triage and Extract for real. If the category is other and an order number is present, it runs Find orders for real. Reply runs for real only when result is a list of one item, and only when the Reply step has a Gmail account. If an account is missing, the action fails. With If it fails set to stop on Reply, the run stops. A published call does not return the sample you see with Send real requests off.

The JSON below is the output this template's unit test expects for the order question when the test does not read the message. The category is billing, Extract returns no order number, and Needs person is simulated. Find orders, Count, One order?, Reply text, and Reply are skipped. This is not a response from a published API call, and it is not the Studio recording above. The recording took the lookup path and still did not reply, because the sample result was {}.

JSON
{
  "request_id": "req_...",
  "run_id": "run_...",
  "workflow": { "id": "wf_...", "version": 1 },
  "status": "succeeded",
  "output": {
    "category": "billing",
    "priority": "low",
    "requires_human_review": false,
    "order_number": null,
    "status_email": false,
    "order_count": null,
    "one_order": null,
    "lookup_result": null,
    "reply_text": null,
    "replied": null,
    "needs_person": true
  },
  "steps": [
    { "id": "input", "status": "succeeded" },
    { "id": "triage", "status": "succeeded" },
    { "id": "extract", "status": "succeeded" },
    { "id": "status", "status": "succeeded" },
    { "id": "find", "status": "skipped", "reason": "branch_not_taken" },
    { "id": "count", "status": "skipped", "reason": "source_skipped" },
    { "id": "one_order", "status": "skipped", "reason": "source_skipped" },
    { "id": "reply_text", "status": "skipped", "reason": "source_skipped" },
    { "id": "reply", "status": "skipped", "reason": "source_skipped" },
    { "id": "needs_person", "status": "succeeded" },
    { "id": "output", "status": "succeeded" }
  ],
  "evaluations": 1
}

The JSON above leaves out credits. A succeeded execute response includes that field. This example is the simulated run, so Needs person is not charged. credits for an API call of this shape is 6 plus Extract data: 5 for Support Triage and 1 for the API run. Extract data is at least 1 credit more, by tokens. The Studio test does not charge the run fee. The recorded test's panel showed 6 credits and 558 Generative AI tokens.

Look up the run later. This returns status, timing, and steps. It does not return the output. Keep the execute response if you need the category or the order number. Runs are kept for 30 days.

cURL
curl https://kovca.io/api/v1/runs/run_... \
  -H "Authorization: Bearer $KOVCA_API_KEY"

18f0a1b2c3d4e5d1 is a key you choose. The message id is a natural one. Keys are kept for 24 hours. A retry with the same key and the same request returns the original run instead of labeling again. A new key can label again. Add label can be repeated without a second effect. Reply cannot: a later run whose count is 1 sends another message. Kovca does not store the message or the output.

To run it from Gmail, open Activate. The checklist is what's left: connect Gmail, connect Shopify, and, once an account is connected, choose that account on the trigger, on Find orders, on Reply, and on Needs person. It says up to 17 credits per event, plus Generative AI by tokens. The activate row is Publish and activate: run on every new email. The button on that row says Publish & activate. That publishes the draft and starts it. A live automation runs that pinned version. Later edits don't reach it until you publish them to the automation.

Connect Gmail and Connect Shopify are still to do. Tested is checked. Up to 17 credits per event, plus Generative AI by tokens. Publish and activate: run on every new email. The button says Publish & activate.

Connect Gmail and Shopify from the checklist, or from the steps, when the routes in the Test tab look right. Until then, leave the accounts empty and keep Send real requests off.

You sign in on Shopify's page. The account line in the Studio says the integration provider holds the store access and uses it for Shopify. Kovca keeps an ID for the account. The template does not store a store. Until Kovca's Shopify app has public distribution, only a development store, or a store the app is already installed on, can connect.

A new message that passes the trigger settings starts a run when Gmail sends the event. Follow runs on the workflow's Executions page (Studio → Usage → Executions). While an automation is paused, events that arrive are skipped. Kovca does not keep them to run later. A failed run is not retried on its own, because the reply may already have been sent.

If someone already answers these emails

You don't need this if a person can read the shared inbox and the volume is small. A saved reply that links to the store's order status page answers the customer who already has that page. A helpdesk that shows the Shopify order beside the thread does the lookup in the place the person already reads mail.

What the workflow adds, on the path this guide demonstrates, is the same classification on every new message that matches the search you set, a copied order number when the email states one, a Shopify lookup on the other path, and a label for a person. Reply stays an optional draft until you have seen a real result. You still choose the accounts, replace to:support@example.com -from:me and Label_needs_person, and read a sample of results before you activate it. The reply text names only the order number from the email. It does not quote fulfillment or tracking.

Make it yours

  • Point the search at the real inbox. to:support@example.com -from:me is a stand-in. Use the search your mailbox actually needs. Set In label once Gmail is connected if the mail already lands under one label, and clear the search if every message in that label should run.
  • Replace the label id. Label_needs_person is not an id from a mailbox. Use the id Gmail lists once the account is connected. The name you see in Gmail is not what Add label sends.
  • Leave Reply off until the lookup shape is known. The sample result is {}. Count does not treat that as one order. After a test on a store you control, point Count at the list you actually see, and rewrite Reply text so it quotes only fields that list contains. Fulfillment status, a tracking number, and a tracking URL are not named outputs of Find orders.
  • Keep the first live replies on the label path. Needs person can run while Reply has no account. Turn Reply on only when a mistaken extra email would be harmless.
  • Do not use limit as a stand-in for "exactly one". Limit is the maximum number of orders on the page (1 to 250). The template leaves it unset. A limit of 1 still returns one row when several orders match.
  • Remember the 60 day window. read_orders covers the last 60 days. An order older than that is not in this lookup.
  • Get order is a different action. Get order (action:shopify/get_order@1) takes an order id, which is not the #1042 name the email states. This template uses Find orders and the name filter. Get order's catalog output is also ok and result, with the same empty sample.
  • Add an answered label only on the reply path. A second Add label after Reply can apply a stand-in id such as one you create for mail this workflow answered. Put it after Reply, on the true branch, so a skipped reply does not apply it. Use a real label id, and leave it off until Reply itself is ready.
  • Change it without surprising a live mailbox. Edit the draft and publish the changes to the automation. The running version stays pinned until you do.

What it costs

Credits, and only for steps that succeed. Billing is the list.

A Studio test skips the run fee. It does not skip Support Triage or Extract. Support Triage is 5 credits when it succeeds, including a test with Send real requests off. Extract is Generative AI, billed by tokens, at least 1 credit, including that test. The recorded test's panel showed 6 credits and 558 Generative AI tokens. A simulated action is not charged, and an action that fails is not charged. A skipped action is not charged. With Send real requests off, Find orders and Needs person return their samples and add nothing. Reply does not run on that sample.

  • A Gmail event: 6 credits for the run.
  • A published API call: 1 credit for the run.
  • Support Triage: 5 credits when it succeeds.
  • Extract: Generative AI, by tokens, at least 1 credit, on every run.
  • Find orders: 2 credits when it runs for real and succeeds. Only the true branch of Has order number? is charged.
  • Reply: 2 credits when it runs for real and succeeds. Only the true branch of One order? is charged.
  • Needs person: 2 credits when it runs for real and succeeds. Only a false branch is charged.

Count, the conditions, and the text step use no credits. A branch that doesn't run isn't charged. The Activate panel adds every step together, including the branch that does not run on the same event, so its number is an upper bound for a live event, and it adds Generative AI by tokens on top. With this template that upper bound is 17 credits per event, plus Extract: the 6 credit event fee, Support Triage, Find orders, Reply, and Needs person. A live event that only applies the label, and does not look up an order, is 13 plus Extract. A live event that looks up an order and labels, and does not reply, is 15 plus Extract. A live event that looks up an order and replies, and does not label, is 15 plus Extract. A published API call is 8 plus Extract when it only labels, 10 plus Extract when it looks up and labels, and 10 plus Extract when it looks up and replies. The response's credits field is the run's total. See the workflows API, HTTP actions, and automations.

On this page

  • What you'll build
  • Watch the test
  • Build it in the Studio
  • Publish, call it, and activate it
  • If someone already answers these emails
  • Make it yours
  • What it costs

Built on Support Triage: Classify inbound support tickets by category and priority, and flag the ones a person should review.

Keep going

Workflows API
InputLead textLead Intellig…Follow up?Search contac…New contact?Create contactOutput
Lead Intelligence24 min read

Add qualified sales emails from Outlook to HubSpot

Build a Kovca workflow that starts from a new Outlook email, classifies it, and creates a HubSpot contact only when no matching contact is found.

Read guide
InputShipment Inte…Needs action?Ticket subjectOpen support …Output
Shipment Intelligence8 min read

Catch delivery exceptions from carrier tracking events

Build a Kovca workflow that reads every carrier tracking update, spots delays, failed deliveries, address problems, and disputes, and opens a support ticket when someone needs to act. Runs from the carrier's webhook. Step by step in the Studio, with a ready-made template.

Read guide
InputSupport TriageHigh priority?Create incide…Alert textPage on-callAdd reactionJoinOutput
Support Triage14 min read

Escalate urgent Slack channel messages into incidents

Build a Kovca workflow that starts from a new Slack message, classifies it, and on high priority opens an incident, posts to on-call, and reacts on the message.

Read guide
Sign inGet Started