RevoChat

Provides a visual interface for non-technical users to construct chatbot conversation flows without writing code, likely using a node-based graph editor or card-based UI pattern where users define intents, responses, and conditional branches. The builder abstracts away NLP complexity by offering pre-built intent templates and slot-filling patterns, then compiles these flows into executable conversation logic that routes user inputs to appropriate response handlers.

Enables one-click or minimal-configuration integration of chatbots into websites via a lightweight JavaScript embed snippet (similar to Intercom or Drift's approach), likely using an iframe or shadow DOM to isolate the chatbot UI from host page styles. The embed script handles authentication, session management, and message routing to RevoChat's backend without requiring developers to modify site architecture or manage CORS complexity.

Allows users to customize the chatbot's appearance to match brand identity, including colors, fonts, logo, and messaging tone. Customization is likely applied through a visual theme editor or configuration panel, affecting the embedded widget's styling without requiring CSS knowledge. The system may support preset themes or allow granular control over individual UI elements (header, message bubbles, input field, etc.).

Provides a catalog of pre-configured conversation flows and intent patterns for common use cases (e.g., FAQ handling, lead qualification, order tracking, appointment scheduling), allowing users to clone and customize templates rather than building from scratch. Templates likely include sample responses, entity extraction patterns, and fallback handling, reducing time-to-deployment and providing best-practice conversation design patterns for non-experts.

Processes user messages through an NLP pipeline to classify intents and extract entities, then routes messages to appropriate response handlers or conversation branches. Likely uses pre-trained language models (possibly fine-tuned on conversation data) or rule-based pattern matching to map user inputs to defined intents, with fallback handling for out-of-scope queries. The routing layer determines whether to respond with a pre-written answer, escalate to a human agent, or trigger an external action.

Maintains conversation state across multiple user messages, tracking variables like user name, previous questions, and conversation history to enable coherent multi-turn interactions. The system likely stores session data in a backend database with TTL-based expiration, allowing the chatbot to reference earlier messages and provide contextually relevant responses. Context is passed to the NLP and response generation layers to inform intent classification and answer selection.

Enables seamless escalation from chatbot to human agents when the bot cannot resolve a query, routing conversations to a queue and notifying available agents through an integrated dashboard or external system. The handoff likely preserves conversation history and context, allowing agents to continue the conversation without requiring users to repeat information. Integration points may include live chat platforms, email, or ticketing systems.

Provides metrics and visualizations on chatbot performance, including conversation volume, intent distribution, user satisfaction, escalation rates, and common unresolved queries. The dashboard likely aggregates conversation logs and extracts insights using basic analytics (counts, averages) and possibly ML-driven analysis (e.g., topic clustering of unresolved queries). Data is presented through charts, tables, and exportable reports to help businesses understand chatbot effectiveness and identify improvement areas.

Enables chatbots to handle conversations in multiple languages, likely through automatic language detection and translation of responses using third-party APIs (Google Translate, DeepL) or built-in language models. Users can define conversation flows in one language and automatically deploy to multiple language variants, or manually create language-specific flows. The system likely stores language preferences per user session and routes messages to appropriate language handlers.

Allows users to define custom variables and slot-filling patterns to collect structured information from users across multiple turns (e.g., name, email, product preference, budget). The system tracks which slots have been filled and prompts for missing information, enabling guided data collection flows. Collected data is stored in session context and can be passed to external systems (CRM, email, etc.) for downstream processing.

Enables chatbots to trigger external actions and integrate with third-party systems through webhooks and REST APIs, allowing conversations to trigger business logic (send email, create ticket, update database, etc.). Users can define webhook payloads and map conversation data to API parameters, enabling the chatbot to perform actions beyond responding with text. The system likely supports request/response handling and error management for failed integrations.