Agent Studio Setup: Build AI Agents in Sparkforge

Agent Studio is a canvas-style builder for orchestrating multiple AI agents and rule-based steps inside one workflow. Each AI Agent node can call LLMs, knowledge bases, web search, and third-party APIs. Sequential nodes handle predictable steps like validation or webhook triggers. The result is a single automation that can reason and act.

What You Get

• Drag-and-drop canvas, no code required.
• Hybrid automation: AI reasoning plus deterministic logic.
• Conditional edges that branch on variables or AI intent.
• Version control with one-click publish and rollback.
• Built-in testing before going live.
• Free tier for unlimited agents; pay only for LLM tokens.

Node Types

AI Agent Node

Connects an LLM prompt to tools like Web Search, Knowledge Base Search, GenAI, and external APIs. Use it for reasoning, content generation, and look-ups.

Sequential Node

Runs rule-based actions: API calls, form validation, webhook triggers. Use it where you need exact, repeatable behavior.

Conditional Routing

Logic lives on the edges, not in separate router blocks. Write expressions like orderTotal > 500 or use AI intent matching to pick the next node. Edges without conditions default to the next step.

Knowledge Bases for RAG

Attach PDFs, CSVs, FAQs, or full websites to any AI Agent node. Retrieval-augmented generation keeps answers grounded in your content. Source-level filters let you scope retrieval per node.

Variables

Store API keys, environment constants, or branding once as Global Variables. Pass runtime values like locationId or userId from the calling app or chat session.

Build Your First Agent

1. From the dashboard, click AI Agents in the left sidebar and select the Agent Studio tab.
2. Click + Create Agent to open the canvas.
3. Drag an AI Agent node from the left panel onto the canvas.
4. Click the node. In the right panel, pick a model (e.g. GPT-4.1) and write a prompt that defines behavior.
5. Drag from the connector dot to wire the node to the next step. Add Sequential nodes for rule-based actions.
6. Optional: open Variables at the top and add globals you want to reuse.
7. Click Test, send a sample input, and review responses, tool usage, and flow behavior.
8. When ready, click Publish > Production.

Pricing

Free tier covers unlimited agents, knowledge bases, and non-AI executions. You pay only for LLM tokens at published API rates.

Version Control and Deployment

Every save creates a draft. Publish to Production with one click, or roll back to any prior version instantly. Test runs use a sandbox quota; only Production runs consume tokens.