CONTENT CREATION

Zero-shot classify R2 images and tag them in Airtable

Runs each image stored in a Cloudflare R2 bucket through a Hugging Face zero-shot classifier against your own label set, then records the top matching tags in an Airtable index.

CategoryContent Creation
Enginesim
Difficultyintermediate
Triggerschedule
Steps6
Setup~15 min

How it runs

The automated pipeline, trigger to output.

  • TriggerScheduled bucket sweep
  • ActionList R2 objects not yet in the Airtable indexCloudflareCloudflare R2
  • ActionGenerate signed URL for each imageCloudflareCloudflare R2
  • ActionZero-shot classify against candidate labels (Hugging Face)Hugging FaceHugging Face
  • LogicKeep labels above confidence threshold
  • OutputUpsert tags into Airtable indexAirtableAirtable

What it does

For every image object in a Cloudflare R2 bucket, this template applies Hugging Face zero-shot image classification using a candidate label list you define (for example product, lifestyle, screenshot, diagram). It keeps the labels that clear a confidence threshold and writes them as tags into an Airtable index row keyed by the object path.

When to use it

You store original media in R2 but need a searchable, human-readable catalog with consistent category tags, and you don't want to train a model. Zero-shot lets you change the label vocabulary anytime without retraining.

How it works

  1. 1A schedule kicks off a sweep of the bucket.
  2. 2List R2 objects and diff against the Airtable index to find untagged images.
  3. 3For each new object, generate a temporary signed URL.
  4. 4Call the Hugging Face zero-shot image classification model with your candidate labels.
  5. 5Filter labels by the confidence threshold and rank them.
  6. 6Upsert the object path and its winning tags into Airtable.

Set it up

What you configure once, before turning it on.

  1. 1
    Connect Cloudflare R2Object storage, S3-compatible.
  2. 2
    Connect Hugging FaceModels, datasets, spaces — the open-source hub.
  3. 3
    Connect AirtableBases, tables, views, automations.
  4. 4
    Set each agent's modelWe leave models unset so you pick the tier — fast + cheap, or top-quality.
  5. 5
    Tune it to your dataEdit the prompts, filters, and field mappings so it matches how your team works.
  6. 6
    Test, then turn it onRun once against a sample, confirm the output, then enable the trigger.

Run this workflow in your colony.

14-day trial. No DevOps. No Sales call. Provisioned in under a minute.