Skip to main content

Every Feature You Need

Radium provides a comprehensive platform for building, deploying, and managing multi-agent workflows. From intelligent orchestration to fine-grained security controls, every feature is designed for production use.

Powerful tools for building and managing complex multi-agent workflows

Intelligent Orchestration

Radium automatically selects the best agent for each task based on capabilities, cost, and performance. No manual routing required—the orchestrator handles agent selection using policy-driven decision making.
[orchestration]
policy = "cost-optimized"  # or "speed-first", "quality-first"
fallback_strategy = "graceful-degradation"

[agent_selection]
prefer_local = true
max_cost_per_call = 0.10
timeout = 30

Multi-Agent DAG Workflows

Define complex workflows as directed acyclic graphs (DAGs) with multiple agents working in parallel or sequence. Built-in dependency resolution, automatic retries, and failure handling ensure reliable execution.
[[workflow.steps]]
id = "analyze"
agent = "code-analyzer"
input = "task.description"

[[workflow.steps]]
id = "generate"
agent = "code-generator"
depends_on = ["analyze"]

[[workflow.steps]]
id = "review"
agent = "code-reviewer"
depends_on = ["generate"]
parallel = ["security-scan"]

Autonomous Execution (YOLO Mode)

Enable agents to run autonomously with minimal human intervention. Agents can make decisions, execute tasks, and handle errors independently. Perfect for long-running tasks or trusted environments.
# Enable autonomous mode
radium-cli chat code-assistant --mode autonomous

# Set execution policy
radium-cli config set execution.mode autonomous
radium-cli config set execution.require_approval false

# Run with bounded autonomy
radium-cli chat code-assistant \
  --mode autonomous \
  --max-iterations 10 \
  --budget 5.00

How Radium Compares

See how Radium stacks up against popular agent frameworks

Feature
Radium
LangChain
AutoGPT
CrewAI
Multi-Agent Orchestration
Coordinate multiple agents in complex workflows
~~
Policy Engine
Fine-grained control over agent behavior
×××
Self-Hosted Models
Run models locally with Ollama
~~
DAG Workflows
Define complex dependencies between tasks
~×~
Cost Tracking
Built-in cost monitoring and budgets
~××
TOML Configuration
Declarative, code-free agent setup
××~
Multiple Interfaces
CLI, TUI, and Desktop app included
×××
Vibe Check Oversight
Metacognitive monitoring of agent behavior
×××
Full Support
~Partial Support
×Not Supported

Ready to Explore?

Try Radium's features yourself or explore our comprehensive documentation to learn more about building autonomous agent workflows.