Skip to content
ClickRuntime

Automate Replicate

Run Replicate models on a schedule and keep track of the jobs. Generate an image or output every morning from the same prompt, hear the moment a long prediction succeeds or fails, and get a daily list of runs that broke. Tell your AI assistant what you need, check what it will do, and it keeps running without you watching the queue.

Add to ChatGPTAdd to Claude
Works inChatGPTClaudeClaude CodeCursorCodexVS CodeGemini CLIZedopencodeAny MCP client

Try asking

Create a ClickRuntime automation: run my upscaling model on Replicate and give it a webhook link. When the prediction finishes, post the status and output link to #ml-jobs in Slack, and tell me if it failed.

Paste it into your assistant. It tells you what it will send or change before it runs live.

Ideas

What you can automate in Replicate

Know when long jobs finish, get fresh model output on a timer, and see failed runs before they cost you more compute. These are ready-made starting points. Anything else you can describe works too.
The moment it happens

Hear when a Replicate job finishes or fails

As soon as a long prediction completes, you get its status and output link in Slack, with no repeat alerts.

See how it works
Daily

Generate fresh model output every morning

Each day a new result from the model and prompt you chose lands in your inbox as a link.

See how it works
Daily

Get a daily list of failed predictions

Each morning you see which runs failed in the past day, which model they used and the error, so you can fix them before rerunning.

See how it works Your own logic

Any other automation you can describe in words

Your own rules, schedule and AI steps with Replicate, plus any other app. Describe it and your assistant builds it.

See how to build your own

Build your own

Build any Replicate automation you can describe

The ideas above are ready-made starting points, not a fixed list. Describe what Replicate should do in your own words, with your conditions, other apps or an AI step, and your assistant writes it, tests it and runs it on ClickRuntime.

  • Your own rules

    Conditions, thresholds, filters and exceptions, written the way you'd explain them to a colleague.

  • Any number of apps

    One automation can read from one app and update several others. Over 248 apps are ready, and anything with an API works too.

  • AI where it helps

    Summaries, classification, extraction and drafts use ClickRuntime's built-in AI, with no extra key.

  • On a schedule or on each event

    Runs as often as every 5 minutes, or the moment an app sends an event, and keeps track of what it already handled.

For example, ask your assistant

Every day, generate a new header image with Replicate from the title of my next scheduled blog post.

For each new row in my Google Sheet of product names, generate an image with Replicate and save it to Google Drive.

When a new Shopify product is added, create lifestyle images with Replicate, upload them to Google Drive and post previews in Slack.

Beyond chat

Have Replicate work when you're not in the chat

Connecting Replicate to ChatGPT or Claude lets you ask about it while you chat. ClickRuntime handles the part that should happen on its own: checks on a schedule and actions on each new event, set up from the same chat and running after you close it.

Setup

Connect Replicate

Replicate connects with a personal API token from your Replicate account. Runs made with it are billed to that account.

Replicate

AI / Model hosting

  1. Sign in at replicate.com and open Account, then API tokens.
  2. Create a token and copy it. It starts with r8_.
  3. In the ClickRuntime dashboard, open Connections & secrets and save the token as a secret for Replicate.

Good to know

Replicate limits that affect automations

  • Replicate bills per second of compute, so a scheduled run costs money every time it fires. Heavier models and longer outputs cost more.
  • Predictions run in the background. An automation either waits for the result or checks back until it's ready, which matters for slow models.
  • Completion webhooks are set on each prediction when it is created, not once for the whole account, so they only cover runs the automation starts.
  • Replicate may resend a webhook, so a finished job can arrive twice. Automations can remember what they've already reported to avoid double alerts.

FAQ

Questions about automating Replicate

Which Replicate models can ClickRuntime run?
Any model your token can run through the Replicate API, using the inputs you describe, such as a prompt, image size or seed. The output links come back to the automation.
Can I get told the moment a prediction finishes?
Yes, for predictions the automation creates. It gives Replicate a webhook link when it starts the run, checks the signature on the callback, and sends you the status and output.
How do I keep an eye on spend?
A daily report of failed runs shows where compute is being wasted. Since billing is per second, running less often or on a smaller model is the most direct way to cut cost.
Do I need a separate AI key?
No. Replicate runs the model itself, and any summary or wording on top of the results uses the AI that comes with every ClickRuntime plan.
Is this a Replicate connector for ChatGPT or Claude?
Not for chatting with Replicate data. A chat connector lets your assistant read and change Replicate while you're in the conversation. ClickRuntime is for the work that should happen without you: you set up an automation from the chat once, and it keeps running on a schedule or on each Replicate event.

Related pages

Keep exploring

Tell your assistant what Replicate should do

Free plan. No credit card.

Add to ChatGPTAdd to Claude