Dialetica AI App User Guide
Complete guide to using the Dialetica AI web application
The Dialetica AI web application provides a complete visual interface for creating, managing, and interacting with multi-agent AI systems. Whether you're building complex agent orchestrations or running simple conversations, the UI offers an intuitive way to work with all platform features.
Agent Management
Create and configure AI agents
Context Workspaces
Multi-agent conversations
Knowledge Base
Manage agent knowledge
Getting Started
After logging in, you'll see your dashboard with quick access to all features:
Main Navigation:
- • Chat: Start new conversations
- • Contexts: Manage conversation workspaces
- • Agents: Create and configure agents
- • Knowledge: Manage knowledge base
- • Tools: Configure MCP integrations
- • Crons: Schedule automated tasks
Account Settings:
- • API Keys: Manage authentication
- • Usage: Track token consumption
- • Account: Profile settings
- • Top-Up: Add credits
Agent Management
Agents are the core building blocks of your AI system. Each agent has unique instructions, capabilities, and configurations.
Agent Creation Steps:
- Click "Create Agent" button on the Agents page
- Enter agent name and description
- Select AI model (GPT, Claude, Gemini, or Auto)
- Configure temperature and max tokens
- Add system instructions (define agent personality and behavior)
- Attach tools (optional MCP integrations)
- Click "Save Agent"
Configuration Options:
- Name: Unique identifier for the agent
- Description: Brief summary of agent's purpose
- Instructions: System prompts that define agent behavior
- Model: Choose from OpenAI, Anthropic, Google, or Auto-routing
- Temperature: Controls randomness (0.0-2.0, default 0.7)
- Max Tokens: Maximum response length
- Tools: MCP tool configurations for extended capabilities
View, edit, and delete your agents from the Agents page.
Agent List Features:
- • View: Click any agent to see full details
- • Edit: Update agent configuration anytime
- • Delete: Remove agents you no longer need
- • Duplicate: Clone existing agents for quick setup
- • Search: Filter agents by name or description
Context Management
Contexts are conversation workspaces where agents collaborate. Create contexts for different use cases like debates, customer support, or research projects.
Context Creation Steps:
- Navigate to Contexts page
- Click "New Context"
- Enter context name and description
- Add system instructions (optional orchestration rules)
- Select agents to participate in this context
- Set visibility (private or public)
- Save your context
The chat interface provides a rich, interactive experience for working with multi-agent contexts.
Chat Features:
Message Sending:
- • Type your message in the input box
- • Use Shift+Enter for new lines
- • Press Enter to send
- • Real-time streaming responses
Agent Interaction:
- • See which agent is responding
- • Watch agents take turns
- • View agent thinking/reasoning
- • Monitor tool usage in real-time
Chat Interface Elements:
- • Message History: Scrollable conversation view
- • Agent Indicators: Visual badges showing active agent
- • Typing Indicators: Real-time feedback when agents are thinking
- • Tool Call Visibility: See when agents use external tools
- • Context Sidebar: Quick access to context settings and agents
Knowledge Management
The Knowledge Base allows agents to access information through semantic search, enabling more informed and accurate responses.
Adding Knowledge Entries:
- Navigate to Knowledge page
- Click "Add Knowledge"
- Enter the knowledge content (facts, instructions, procedures)
- Choose scope: User-level, Context-specific, or Agent-specific
- Add metadata tags for organization (optional)
- Save the entry
Aditionally, you can add knowledge entries from a message (or a selection of text) by clicking the "Add to Knowledge" button in the message box.
Knowledge Scope Options:
- • User-Level: Available across all your contexts and agents
- • Context-Level: Available only within a specific context
- • Agent-Level: Exclusive to a single agent
All knowledge entries are automatically embedded and can be searched semantically by agents during conversations.
- • Automatic Retrieval: Agents search knowledge based on conversation context
- • Semantic Matching: Finds relevant information even with different wording
- • Manual Search: Test queries in the Knowledge page
- • Relevance Ranking: Most relevant entries appear first
MCP Tools Configuration
MCP (Model Context Protocol) tools extend agent capabilities by connecting them to external services like Notion, GitHub, databases, and custom APIs.
Setting Up Tool Configurations:
- Go to Tools page
- Click "Add Tool Configuration"
- Enter tool name and description
- Provide MCP server endpoint URL
- Add authentication token (if required)
- Select connection type (HTTP or SSE)
- Save configuration
- Attach to agents that need this tool
Popular MCP Integrations:
- • Notion: Read and write Notion pages and databases
- • GitHub: Access repositories, create issues, review code
- • Slack: Send messages, read channels
- • Web Search: Perform real-time web searches
- • Custom APIs: Connect any MCP-compatible service
You can find a list of useful servers at mcpservers.org. For straightforward and unified integration with Google, Notion, Zoom and others, we recommend waystation.ai.
Once configured, tools can be attached to agents to extend their capabilities.
- • Open Agents settings
- • Hover over the agent you want to configure, and click the "Edit" button
- • Navigate to the "Tools" section
- • Select tool configurations from your list
- • Save changes
- • Agent can now use these tools in conversations
Scheduled Tasks (Cron Jobs)
Schedule automated tasks that execute prompts within contexts at specific times or intervals.
Setting Up Cron Jobs:
- Navigate to "Crons" page
- Click "Create Cron Job"
- Enter job name and description
- Write the prompt to execute
- Select target context
- Choose schedule type:
- - Recurring: Use cron expression (e.g., "0 9 * * *" for daily at 9 AM)
- - One-time: Select specific date and time
- Save and activate the cron job
Common Cron Expressions:
- • 0 * * * * - Every hour
- • 0 9 * * * - Daily at 9 AM
- • 0 9 * * 1 - Every Monday at 9 AM
- • 0 0 1 * * - First day of every month
- • */15 * * * * - Every 15 minutes
Track the status and execution history of your scheduled tasks.
- • Status Indicators: See if jobs are active, paused, or failed
- • Last Run: View when the job last executed
- • Next Run: See upcoming execution time
- • Execution History: Review past runs and results
- • Pause/Resume: Temporarily disable jobs without deleting
Settings & Account Management
Generate and manage API keys for programmatic access to Dialetica AI.
Creating API Keys:
- Go to API Keys page
- Click "Create New API Key"
- Enter a descriptive name
- Copy the generated key (shown only once)
- Store securely - use in your applications
⚠️ Security Note:
API keys grant full access to your account. Never share them or commit them to version control. Rotate keys regularly and delete unused ones.
Monitor your token consumption, API usage, and costs in the Usage dashboard.
- • Token Usage: Track input and output tokens
- • Cost Breakdown: See spending by model and capability
- • Request History: View API call logs
- • Usage Trends: Analyze consumption patterns over time
- • Export Reports: Download usage data for analysis
Manage your profile, preferences, and account details.
- • Profile: Update name, email, and avatar
- • Password: Change account password
- • Preferences: Set default models, themes, and notifications
- • Billing: Manage payment methods and view invoices
- • Credits: Top up account balance for API usage
Tips & Best Practices
- • Give agents clear, specific instructions
- • Start with lower temperature (0.3-0.5) for factual tasks
- • Use higher temperature (0.7-1.0) for creative tasks
- • Test agents individually before adding to contexts
- • Name agents descriptively to clarify their roles
- • Create separate contexts for different projects
- • Add 2-4 agents per context for best results
- • Use descriptive names for easy identification
- • Set clear instructions for agent collaboration
- • Archive completed contexts to reduce clutter
- • Keep knowledge entries focused and concise
- • Use metadata tags for better organization
- • Regularly update outdated information
- • Test semantic search to verify relevance
- • Organize by scope (user/context/agent level)
- • Use Auto Mode for cost-effective operations
- • Set appropriate max_tokens to control costs
- • Enable streaming for better user experience
- • Monitor usage dashboard regularly
- • Clean up unused agents and contexts
Explore our comprehensive documentation for detailed information on API usage, SDK integration, and advanced features.