Studio and Workflow Creation
Studio
Easily design workflows with our no-code drag-and-drop builder, and connect multiple integrations in one place.
Add Nodes
Start building your workflow here by adding nodes.
Input Prompt Node
Enter your prompt here in plain text or JSON. You can type it directly, pull it from a previous node, or combine it with other prompts to provide context for the next node.
Best practices:
Use structured JSON prompts for consistency when passing data between nodes.
Keep prompts modular and reusable by storing common instructions separately (e.g., tone, style, constraints).
When chaining prompts, label variables clearly (e.g.,
{{customer_name}},{{invoice_amount}}) so they can be reused downstream.Validate and sanitize any user-generated input before feeding it in to avoid workflow errors or unexpected outputs.
Example:
You’re building a workflow to generate a Bitcoin treasury market update.
In plain text: "Summarize Bitcoin price movements in the past 24 hours"
In JSON:
This prompt could be fed downstream into a language model node to generate a CFO-ready report.
Render Output Node
This node displays the output from the previous step, including text, images, or videos. It can also be used for debugging—letting you check results at each stage—or to pass content forward as input to the next node.
Best Practices:
Use render nodes early and often during workflow design to debug data flow.
Add a render node before complex branches (e.g., conditionals, loops) to ensure inputs are correct.
Avoid cluttering production workflows with too many render nodes—keep them in dev/test versions for troubleshooting.
For media outputs, ensure file sizes are manageable to avoid slowing down the workflow.
Example:
You want to debug whether your Bitcoin price feed is returning data properly.
The Render Output node displays:
This helps you confirm the API is working before passing it into your reporting workflow.
Workflows Node
Saved workflows can be reused as sub-workflow nodes. Select any existing workflow from the dropdown to include it in the one you’re building.
Best Practices:
Think of writing workflow descriptions like writing prompts. Include context, intent, and outcomes so agents can choose the right workflow when multiple options are available.
Break complex automations into modular sub-workflows to improve reusability.
Name workflows clearly with a verb + object format (e.g., “Validate Customer Data” instead of “DataCheck1”).
Use sub-workflows for repeatable tasks (e.g., formatting data, sending notifications).
Version your workflows (e.g.,
v1,v2) when making changes so you can roll back if needed.
Example:
You’ve already built a workflow called “BTC Price Fetcher” that pulls Bitcoin prices from multiple exchanges and averages them.
Inside your Treasury Stress Test Workflow, you drop in the “BTC Price Fetcher” sub-workflow instead of rebuilding it.
This ensures that any downstream risk analysis always works off standardized BTC price data.
Loop Process Node
This node processes an array by sending each item one at a time into the next node. The workflow runs to completion for each item before moving to the next.
Best Practices:
Use loops only when necessary; for large datasets consider batching items to improve performance.
Include error handling inside loops so that if one item fails, it shouldn’t break the whole process.
When possible, use render output node to log or render the current item during testing to confirm correct iteration.
Example:
You need to evaluate multiple Bitcoin transactions for compliance.
Input array:
Loop Process node feeds each transaction one by one into a compliance check sub-workflow (e.g., “Verify KYC for sender”).
Conditional Node
The conditional node compares two strings and if they match, the output goes to the true branch and triggers the part of the workflow after this branch. Otherwise, the output goes to the false branch, triggering the part of the workflow after this branch.
Best Practices:
Normalize data (trim spaces, lowercase, etc.) before comparing strings to avoid false negatives.
Keep conditions simple and explicit. Avoid nesting too many conditionals as it makes workflows harder to debug.
When conditions become complex, consider offloading logic to a code node or external function.
Example:
You want to verify that a transaction is confirmed before posting it to the ledger.
Comparison: {{tx.status}} equals "confirmed"
True branch → Continue to “Record Transaction in Ledger”
False branch → Route to “Await Confirmation”
Agents Node
Agents created on the Agents page can be added as nodes in any workflow. This lets you reuse specialized agents without recreating them each time.
Best Practices:
One agent, one task. Use agents for domain-specific tasks (e.g., summarization, compliance checks, financial analysis). Mixing too many tasks may confuse the models.
Keep agent prompts consistent across workflows—document what each agent does and where it’s best used.
Reuse existing agents instead of duplicating—this reduces maintenance.
Test agents in isolation before embedding them into production workflows.
Example:
You’ve created an “On-Chain Analytics Agent” that monitors large BTC wallet movements.
In your workflow:
Agent is invoked to summarize “Top 3 whale movements in the past 24h.”
Output feeds into your daily Bitcoin Market Risk Report.
Integrations Nodes
Here you can find all the integrations that are available on Orchestrator. These integrations can be used as nodes in your workflows.
Unlinked services display an Add connection option; once authorized, the integration is provisioned for organizational use.
Best Practices:
Regularly rotate API keys or tokens to maintain security.
Test connections in a sandbox/staging account before applying them to production services.
Document which workflows depend on which integrations to simplify troubleshooting later.
Example:
Connect to Strike to automate Bitcoin payments.
Integration node: “Send BTC Payment via Strike.”
Input fields: recipient’s Lightning address, amount in sats, optional memo.
Workflow result: once triggered, the node executes a Bitcoin payment over Lightning through Strike’s API and returns the payment status.
HTTP Request Node
You can create HTTP request resources to call any external services with an API, then add them to workflows as nodes. This is useful if you wish to leverage an API but we’ve yet to offer an integration.
To create a new HTTP request resource, click on HTTP Requests > Create new HTTP Request, then enter the name, method (GET, POST, PUT, DELETE, PATCH) and URL. The header and payload is optional.
Once you create a HTTP request resource, you can then use it in any workflows available in your organization as a node.
Best Practices:
Always test API calls in Postman or curl first before embedding them into workflows.
Add retry logic for unreliable APIs to handle timeouts or rate limits.
Keep requests modular. Create one HTTP resource per distinct endpoint instead of overloading a single one with multiple use cases.
Log API responses during development using render output node to confirm the data structure matches your expectations.
Example:
You need Bitcoin mempool data for fee estimation.
Create HTTP request resource:
Name: BTC Mempool Fee Rates
Method: GET
URL:
https://mempool.space/api/v1/fees/recommended
Example response:
This feeds into your “Transaction Processor Workflow” so outgoing BTC payments always include optimal fees.
Last updated