AI Pipeline Builder

The AI pipeline builder lets you describe a workflow in plain English and Wrapd automatically maps it to a sequence of your existing endpoints. Instead of manually dragging nodes and connecting edges, you describe what you want and the AI assembles the pipeline for you.

How to use it

1. Go to Dashboard → Pipelines and click New Pipeline
2. Click Generate with AI
3. Describe your workflow — for example, “back up the database, then compress the dump, then upload it to S3”
4. The AI analyzes your existing endpoints and maps each step to a matching endpoint by slug and description
5. Review the generated pipeline in the visual editor
6. Adjust any steps, add condition nodes, or tweak parameters as needed
7. Click Save

You can refine the result by typing follow-up instructions like “add an error notification step at the end” or “skip the upload if the backup is empty.”

How AI matches steps to endpoints

The builder looks at all endpoints registered to your account and matches each step in your description to the most relevant endpoint based on:

• Slug — exact or partial matches (e.g., “back up the database” matches db-backup)
• Description — semantic similarity to what you described
• Parameters — the AI pre-fills parameter mappings and capture references between steps

If the AI uses an endpoint, it configures the step node with the correct slug, parameters, and data-passing references automatically.

When no matching endpoint exists

If the AI cannot find an endpoint that matches a step in your workflow, it flags the step with a warning. You will see:

• A highlighted node in the visual editor indicating no endpoint match
• A suggestion to create the missing endpoint first (with a link to the endpoint creation form)

The pipeline can still be saved with unmatched steps, but those steps will fail at execution time until a matching endpoint is created.

Tips for better results

• Be specific about the sequence — “first do X, then Y, then Z” works better than “do X, Y, and Z”
• Use endpoint names when you know them — “run db-backup then compress-file” gives exact matches
• Describe conditions explicitly — “if the backup size is greater than 0, upload it; otherwise send an alert”
• Mention data flow — “pass the output filename from step 1 to step 2” helps the AI set up capture references
• Keep it to 3-8 steps — very long workflows are better built in the visual editor directly

Examples

Prompt	What you get
”backup postgres, compress, upload to s3”	3-step linear pipeline with data passing
”check disk usage, alert if over 90%“	2 steps with a condition node for the threshold
”run tests, deploy to staging if they pass”	2 steps with a condition node on exit status
”fetch latest logs, grep for errors, email report”	3-step pipeline with capture patterns

Tier requirements

AI pipeline generation is available on Pro and Team plans.

Plan	AI pipeline generations / day
Free	0
Pro	10
Team	50

Credits reset daily. Your remaining credits are shown in the AI generation modal.