AI‑Powered Workflows
Learn how to integrate local LLMs for deeper content analysis.
This tutorial builds on the Quick Intro and shows a slightly more “real” workflow.
You’ll create an automation that:
As before, everything runs entirely in your browser — no server required.
graph TB
A[Web Page] --> B[Get All Links]
B --> C[Get All Text]
C --> D[Edit Fields - Analysis]
D --> E[If Node - Logic]
E -->|Long Article| F[Detailed Report]
E -->|Short Article| G[Quick Summary]
This shows the main idea behind workflows: they can react to what they find on the page.
Agentic WorkFlow.<a> link URLs and labels from the current page.You’ll see the list of links in the node output.
This shows how to gather all hyperlinks on the page.
From Get All Links, click Add node.
Search for Get All Text and add it.
Run Execute Step to fetch the full page text.
From Get All Text, click Add node again.
Insert Edit Fields to analyze the text.
Configure these fields by copying the values below:
{{ $json.text.split(' ').length }}{{ $json.text.length }}{{ $json.text.substring(0, 200) }}...Execute to see analysis results.
This gives you a quick “profile” of the page: how long it is, plus a short preview.
graph TB Analysis[Analysis Output] --> Branch[If Node] Branch -->|word_count > 500| Long[Detailed Branch] Branch -->|otherwise| Short[Summary Branch]
From the Edit Fields node, click Add node.
Search and insert If.
Configure it:
word_count (select this from the previous node’s output)500Execute the If node. You’ll see two potential paths: “true” (longer than 500 words) and “false” (shorter).
Now your workflow can take different paths based on content length.
For Long (true) branch:
report_type: “Detailed Analysis”
message:
Long article: {{ $json.word_count }} words, {{ $json.character_count }} charsPreview: {{ $json.summary }}For Short (false) branch:
report_type: “Quick Summary”
message:
Short article: {{ $json.word_count }} wordsPreview: {{ $json.summary }}Then execute the full workflow. The final output will differ based on whether the content was “long” or “short.”
Navigate to a web page with substantial text (e.g. a news article).
In the builder, click Execute Workflow (runs from first node onward).
Watch as the workflow runs step-by-step:
Inspect the output of the final node to see the generated message.
You now have a branching, content‑aware automation built entirely in the browser.
AI‑Powered Workflows
Learn how to integrate local LLMs for deeper content analysis.
Browser Nodes
Explore all supported browser manipulation nodes.
Key Concepts
Review the fundamentals of triggers, nodes, and data flow.
Troubleshooting & Best Practices
Tips to debug, optimize, and make workflows more robust.