> ## Documentation Index > Fetch the complete documentation index at: https://docs.diga.io/llms.txt > Use this file to discover all available pages before exploring further. # Updates > Release notes and platform improvements for Diga ## Pre-call flows and flow-driven variables ### Pre-call flows * **Prepare every call before it starts**: the new **Pre-Call** trigger runs a workflow the moment a call begins, before the agent greets. The agent waits for the result and uses it to personalize the conversation. * **Return Call Data action**: send a prompt addition and dynamic variables back to the agent, so it can start the call already knowing the caller's name, account, or context. * **Gate calls**: optionally abort a call if the flow fails or times out — useful for allowlists, eligibility checks, or business hours. Timeout is configurable per flow (up to 55 seconds). * **Assign anywhere**: add pre-call flows to single-prompt agents in the Workflows section, or to conversational paths through the new **Pre-call flows** node above the start node. ### Flow-driven dynamic variables * **Variables provided by flows**: variables a pre-call flow produces now appear in the agent's dynamic variables panel as read-only, labeled "Provided by \[flow]", and in the `{{` autocomplete. * **Dynamic parameters for during-call flows**: a Function Call parameter can now take its value from a dynamic variable instead of being extracted by the agent — set its **Value Source** to **Dynamic Variable**. * **Variables in paths**: dynamic variables now interpolate in conversation node prompts and branch conditions. ### Workflow categories * **New "Call launcher" category**: workflows that start outbound calls with the Make Call action are now labeled **Call launcher**, distinct from the new **Pre-call** flows. ## Improved voices, tool sounds, and Holded piece ### More control over voices * **Expanded voice catalog**: Browse thousands of available voices and choose the best fit by language, accent, tone, and more * **Direct ElevenLabs support**: Use any ElevenLabs voice by entering its ID directly * **New voice controls**: Adjust voice temperature and volume, in addition to the existing speed control ### Sound in tools and flows * **Background sound during execution**: Add sound to tools or flows while they run so calls sound more natural * **Pre-execution announcement**: Force the model to say something before executing a tool, so it always announces what it's about to do ### New piece: Holded * **Holded integration**: Connect your agents with Holded to talk directly to your ERP — for example, assist customers by looking up their invoices in real time ## Improved turn detection model * **Better turn detection**: Improved model with GPU support for more accurate detection of when the user has finished speaking * **Background sound in flows**: Tools and flows can now play background sound while executing ## Workflow and session fixes * **Workflow fixes**: Stability improvements in the workflow builder * **OAuth session persistence**: Fixed an issue where Google OAuth sessions were not persisting across tabs, even with the "Remember me" option enabled ## Bulk recording downloads and RAG improvements * **Bulk recording downloads**: Download multiple call recordings at once from the monitoring dashboard * **RAG improvements**: Bug fixes and improvements to the knowledge base for more reliable information retrieval ## Permission and navigation fixes * **Permission loading**: Fixed an issue with permission loading on login * **Sidebar visibility**: Fixed menu item visibility based on user permissions * **404 page**: Improved error page for unknown routes ## Logs, RAG, and Workflows ### Call Logs * **Detailed event logging**: Every call generates structured logs covering its full lifecycle — from connection to completion * **10 event categories**: Call lifecycle, user messages, agent responses, call tools, integrations, conversational paths, transfers, and webhooks * **Customer-facing logs**: Dual logging system that automatically filters out internal technical data and exposes only relevant events to users via the API * **Full traceability**: Each event includes timestamps, trace IDs, and contextual data to debug and analyze agent behavior ### RAG (Retrieval-Augmented Generation) * **Knowledge base for agents**: Upload documents and URLs so your agents can access relevant information during conversations * **Semantic search**: The system automatically indexes and searches your content to provide accurate, contextual responses * **Multiple sources**: Ingest files, URLs, and text, with automatic page discovery from a domain * **Knowledge management**: Add, update, and delete items from your agents' knowledge base ### Workflows * **Visual automation**: Create no-code workflows that run during or after calls * **Post-call workflows**: Automate tasks after a call ends, such as updating a CRM, sending emails, or logging data * **During-call workflows**: Allow your agent to execute tools in real time, like scheduling appointments or looking up data * **Sync and async execution**: Choose whether the agent waits for the workflow response or continues the conversation * **Pre-built templates**: Use ready-made templates to quickly set up common workflows ## Diga Initial Launch We're excited to introduce Diga, the complete platform for creating and deploying AI-powered voice agents. ### Agent Building * **Single Prompts**: Create conversational agents with simple and direct prompts * **Conversational Paths**: Define complex conversation flows with custom nodes and routes * Support for multiple node types (message, question, decision, action) * Visual editor to design the conversational flow * **Call Customization**: Configure call behavior and characteristics * Customizable voices * Language and tone configuration * Speed and volume parameters * **Version Control**: Safely manage different versions of your agents * Complete change history * Rollback to previous versions * Version comparison * **Call Tools**: Add special capabilities during conversations * Call transfer * Send messages * Execute custom actions ### Integrations * **HTTP Integrations**: Connect your agents with external APIs * Inbound and outbound webhooks * Flexible authentication (API Key, Bearer Token, OAuth) * Custom data mapping * **MCP Integrations**: Model Context Protocol support * Native integration with MCP tools * Extensibility for advanced use cases * **Add Integrations to Agents**: Modular system to connect multiple integrations ### Telephony * **Phone Number Purchase**: Acquire numbers directly from the platform * Availability in multiple countries * Search by area code * Immediate activation * **SIP Trunks**: Configure and manage your own SIP trunks * Inbound and outbound calls * Routing configuration * Multi-provider support ### Platform Management * **Projects**: Organize your agents and resources * Isolated workspaces * **Billing**: Detailed cost control * Flexible plans * Real-time consumption monitoring * Usage reports * Configurable limits ### Analytics and Monitoring * **Monitoring**: Complete visibility into your agents' behavior * Complete transcriptions ### Documentation and Tools * **Complete API Reference**: Interactive documentation with OpenAPI * Examples in multiple languages (Python, JavaScript, cURL, Go) * Interactive playground * Authentication and code examples * **Documentation in Spanish**: Complete guides and tutorials * **Documentation in English**: English version available