⚙️ How to Create a Workflow
This guide will walk you through building, editing, and running workflows in Twazi. Workflows are made by connecting tasks together to automate complex processes involving AI, blockchain, and web data.
Step-by-Step: Example Workflow
Goal: Create a workflow that:
- Creates a recall bucket
- Calls a Large Language Model (LLM)
- Stores the LLM reply in the recall bucket
Prerequisites
-
SambaNova Cloud API Key
- Create one at SambaNova Cloud.
-
Recall Funded Account
- Use the Recall Faucet.
- Learn more at Recall Network.
⚠️ Note: You must add your credentials in the Credentials section of the Twazi dashboard before creating your workflow.
1. Create a New Workflow
- Go to the Workflows section in the Twazi dashboard.
- Click Create Workflow.
- Give your workflow a name and description.
2. Add Tasks to the Workflow
a. Create Recall Bucket
- Add the Create Recall Bucket task.
- Select your Recall private key credential.
b. Get Reasoning from LLM
- Add the Get Reasoning from LLM task.
- Select your SambaNova Cloud API key credential.
- Provide the messages or prompt you want the LLM to process.
- Optionally reference the bucket address to store context.
c. Store Streamed AI Output
- Add the Store Streamed AI Output task.
- Set the Bucket Address to the first task’s output.
- Set the Full AI Output to the LLM reply.
- Select your Recall private key credential again.
3. Connect the Tasks
- Drag to connect the output of each task to the input of the next.
- Ensure data flows correctly: Bucket → LLM → Storage.
4. Save and Run
- Click Save to store your workflow.
- Click Run to execute.
- Monitor results using the Execution Monitor.
Tips
- You can edit any workflow by selecting it and clicking Edit
- Use the Execution Monitor to debug and inspect.
- Credits are consumed based on task usage—check your balance!
🔁 You now know how to create a full Twazi workflow using AI, blockchain, and decentralized storage—all without writing a line of code.
See more details in Workflow Tasks.