DeepScript

Integration

Make.com Integration — audio pipelines with branching and iteration

Drive watcher, transcription, GPT summary, email digest — all in one visual scenario.

Make.com (formerly Integromat) is Zapier's more powerful sibling: branches, iterators, aggregators, error handlers and rollbacks. Our DeepScript module ships `Upload Audio`, `Watch Transcriptions` (polling trigger), `Watch Webhooks` (instant trigger), `Get Transcript`, `Export Transcript` and `Search Transcripts`. Unlike Zapier, Make charges per operation rather than per task — complex scenarios with many steps end up significantly cheaper. Typical pattern: a Google Drive watcher detects new audio files, DeepScript transcribes, OpenAI summarises, an iterator splits by speaker, and a formatted Markdown digest lands in your inbox.

View on the Make App Hub

What you can build

  • Watch a Drive folder → transcribe → summarise with Claude → store in Notion.
  • Iterate over speaker segments: each speaker gets their own card in Trello/Linear.
  • Error handler with fallback: on `transcription.failed`, auto-create a support ticket.
  • Aggregator: collect all transcriptions of a day and post them as a Slack digest at 6pm.
  • Branched logic: audio < 10 min → Standard model; > 10 min → Premium with vocabulary.

Code samples

Module overview (DeepScript app on Make)JSON
{
  "app": "deepscript",
  "modules": [
    {
      "name": "Upload Audio",
      "type": "action",
      "operations": 1,
      "inputs": ["file", "model", "language", "vocabularyId"]
    },
    {
      "name": "Watch Transcriptions",
      "type": "polling-trigger",
      "interval": "5min",
      "operations": 1
    },
    {
      "name": "Watch Webhooks",
      "type": "instant-trigger",
      "operations": 1,
      "events": ["transcription.completed", "transcription.failed", "balance.low"]
    },
    {
      "name": "Get Transcript",
      "type": "action",
      "operations": 1,
      "inputs": ["transcriptionId", "includeWords"]
    },
    {
      "name": "Export Transcript",
      "type": "action",
      "operations": 1,
      "inputs": ["transcriptionId", "format"]
    },
    {
      "name": "Search Transcripts",
      "type": "action",
      "operations": 1,
      "inputs": ["query", "limit"]
    }
  ]
}
Scenario: Drive → Transcribe → Claude → EmailYAML
scenario:
  name: Daily Meeting Digest

  trigger:
    module: google-drive.watch-files
    folder: /Meetings/Today

  steps:
    - id: 1
      module: deepscript.upload-audio
      file: "{{trigger.file}}"
      model: premium
      language: de

    - id: 2
      module: deepscript.watch-webhooks
      event: transcription.completed
      filter: "{{step1.id == webhook.data.transcriptionId}}"

    - id: 3
      module: deepscript.get-transcript
      transcriptionId: "{{step2.data.transcriptionId}}"

    - id: 4
      module: anthropic.create-message
      model: claude-sonnet-4-5
      prompt: |
        Summarise the meeting transcript in 5 bullet points,
        extract action items, and list speakers.
        Transcript: {{step3.text}}

    - id: 5
      module: gmail.send-email
      to: team@example.com
      subject: "Meeting Digest: {{trigger.file.name}}"
      body: "{{step4.response}}"

  errorHandler:
    on: step1.failed
    module: gmail.send-email
    to: ops@example.com
    subject: "DeepScript upload failed"

Setup in a few steps

  1. 1

    Create a DeepScript connection

    In the Make editor, add the DeepScript module, choose "Add Connection" and paste the API key from your dashboard. Make stores the connection encrypted and you can reuse it across any number of scenarios.

  2. 2

    Pick a trigger — polling vs webhook

    `Watch Transcriptions` polls every 5-15 minutes (one operation per poll) — fine for small scenarios. `Watch Webhooks` is instant and only costs 1 operation per actual event — recommended for production.

  3. 3

    Wire up the processing logic

    Drag and drop modules to connect them. Use routes for branches, iterators for lists and aggregators to collect data. DeepScript output fields (text, words, duration, speakerCount, cost) are immediately available in downstream modules.

  4. 4

    Add error handlers

    Make lets you attach an error handler per module. Tip: on DeepScript modules, configure a `Resume` handler with a 30-second wait — that absorbs transient 5xx errors without aborting the scenario.

Frequently asked questions

How many operations does a typical scenario use?

The sample workflow (Drive → Upload → Webhook → Get → Claude → Mail) costs 5 operations per audio file. At 100 meetings/month that's 500 ops — comfortably inside Make's Free plan (1000 ops). DeepScript API costs come on top (€0.18 or €0.27 per hour of audio).

Which trigger is cheaper — polling or webhook?

Webhook almost always. Polling every 5 min = 288 operations/day, even if nothing happens. Webhook = 1 operation per actual event. Exception: if you want to pick up DeepScript transcriptions created from third-party sources (manual, external tools) and the trigger only watches your account.

Can I version-control Make scenarios?

Make has a built-in versioning system (Teams/Enterprise plans). Alternatively: export scenarios as JSON and commit them to Git. DeepScript connections aren't included in the export — they must be created per workspace.

What happens during a Make outage?

DeepScript jobs keep running independently — they sit in our queue. Once Make is back online the webhook trigger replays missed events (we retry webhooks 3 times with backoff). Polling triggers may miss events, which is another reason to prefer webhooks.

Does the Make integration differ from Zapier?

Functionally identical (same API underneath). Make is cheaper per operation, has more powerful flow logic (routes, iterators, aggregators) and a more visual editor. Zapier has the larger app directory. For complex pipelines, Make is the better fit.

Ready to ship this to production?

Create an account, generate an API key, ship. Three transcriptions free to try. Full OpenAPI 3.1 docs at api.deepscript.com/docs.

Make.com Integration — complex audio workflows | DeepScript