6 posts tagged with "nAIxus"

🚀 Features​

✨ Feedback Campaign Management​

We are thrilled to introduce a new capability in nAIxus: Feedback Campaign Management, empowering administrators to seamlessly collect user insights throughout any intelligent assistant scenario. Administrators can now create and manage feedback campaigns directly within the nAIxus console to capture Customer Satisfaction Score (CSAT) ratings. This marks the first step in a broader feedback strategy (NPS and binary feedback types coming soon).

Key Functionalities​

• Campaign Lifecycle:

  A feedback campaign can be in one of three states:

  • Draft – editable, but not yet active
  • Published – active and available for feedback collection
  • Closed – archived and no longer accepting responses

• Embedded Feedback Collection:

  Via a new dedicated step-block, administrators can prompt users for feedback at any point in a process, by linking to a published campaign.

• Campaign Analytics Dashboard:

  Track:

  • Number of feedback requests sent and responses received
  • Response breakdown (score distribution from 1 to 5)
  • Real-time CSAT score
  • Anonymous comments with timestamps
  • Scenario traceability: where and when feedback was collected

Smart Design, Future-Ready​

• Fully integrated into scenario flows via drag-and-drop blocks
• Alerts guide the admin if an unpublished campaign is selected
• Designed with extensibility in mind: future updates will support timed expirations, additional feedback types (NPS, binary), and enhanced data visualizations (e.g. average score heatmaps per scenario)

This new feature reinforces nAIxus' mission to combine predictive, generative, and soon agent-based AI with robust user-centric insights—allowing your assistant to learn, adapt, and improve continuously.

Let us know how you feel about it 😉

✨ NLP Training Import Feature Update​

We are pleased to announce enhancements to the NLP Training Import functionality in our latest release:

Improvements:​

• Enhanced validation checks for imported training phrases, providing clearer, actionable feedback to users.
• Multilingual support now distinctly organized by classifier enabled locales, facilitating easier management of language-specific intents.

Validation Enhancements:​

• Duplicate Detection: Training phrases existing in multiple intents are now clearly flagged.
• Whitespace Validation: The system now prevents importing training phrases composed solely of whitespace characters.
• Scenario Validation: Users are clearly informed if a referenced scenario does not exist in the specified language.

We recommend reviewing your NLP data following import to ensure accuracy. As always, remember to retrain your virtual assistant after modifying the NLP training data.

✨ Omnichannel Updates​

Webhook Channel: API guide documentation is now directly available within the "How to Use" tab, enhancing ease of integration. Please note the documentation is only available in English.

🚀 Features​

✨ Agents​

We are welcoming agents to our families of Blocks. We will start with the great Retrieval Augmented Generation block, allowing your to search among any Knowledge Base created in the Knowledge Module.

We hope to later add Reasoning Agents, Redacting Agents, Summarizing Agents, Free-Prompting Agents and much more…

✨ Transcripts Renewed​

We're excited to introduce the newly redesigned interface, delivering an improved and more intuitive user experience!

What's New:​

• Enhanced Historical View: Conversations are clearly organized by date and time, with quick visual indicators showing conversation status, platform used (web/mobile), and service type.
• Interactive Filtering Options: New user-friendly filters allow quick selection of languages, date ranges and release mode, simplifying historical conversation retrieval and analysis.
• Improved Conversation View: Clear visual cues for conversation flow and assistant interactions. Easily identifiable timestamps and detailed interaction indicators.

We look forward to your feedback as you explore this new experience!

✨ Enhanced Webchannel UI​

We've updated the Webchannel by removing the “block effect” around assistant messages, creating a cleaner, more modern user experience.

Additional personalization enhancements will follow in a future release.

🚀 Features​

✨ New Editor​

We reviewed our editor, which to be honest was horrible for us to manage. Easier management, more features, what’s not to ask for ? We added great functionnalities along the way, just because 😎

• Quick action menu on text selection
• Quick action menu on new empty line
• Adding a cursor when drag and dropping element from within the editor
• Adding image from a URL
• History management : undo and redo for ever !
• Lists : ordered, unordered, task

🚀 Features​

✨ Condition Block​

• New Action Block Type allowing scenario designers to make decisions based on context variables values

✨ Publish​

Environment variables​

You can now create new environment variables that are common to all scenarios. Those variables can be used freely in blocks.

Versioning​

A tenant now allows for two environments to coexists allowing users to safely work on their scenarios without altering production capabilities :

• DEVELOPMENT which corresponds to your current work environment. All your actual compatible (see below) components will be marked as
• PRODUCTION which is mostly read-only and so is isolated from changes made in development as it can only be modified through a publish action.

Versioned components are:

• Scenarios
• NLP Configuration
• LLM Configuration
• Anonymizations
• Smalltalk
• Channels
• Environment Variables

Publish​

• Publishing is the action of deploying your dev environment in production and making it accessible for your production channels.
• It will replace all your PRODUCTION components by what is in DEVELOPMENT. (Scenarios needs to be on State READY as well to be published). Except for Environment Variables that need to be modified manually every time.
• It is an destructive action, meaning that once the publish is started, it may not be possible to revert back to a previous state. Proceed with caution.

🚀 Features​

✨ Statistics​

• Can see which scenario have been triggered on a selected timeframe. Not retroactive, it will only starts collecting data from new version onwards

✨ Entities​

Detection​

• Users can use “Ask Entity” blocks in which they can ask users for certain entities. Recognition is done using IA for NER (named-entity recognition) or by patterns using a list or a regex and store entities extracted from the answer in a variable.
• Available entities are:
  • List
  • Regex
  • Date (NER)
  • Time (NER)
  • Phone (NER)
  • Number (NER)
  • Currency (NER)
  • Email (NER)
  • URL (NER)
  • Name (NER)
  • Address (NER)
  • City (NER)
  • Country (NER)
  • Location (NER)
  • Timezone (NER)
  • Language (NER)

Anonymization​

• Users can define entities to be anonymized and choose values to replace those in three independent cases: RAG calls, classifier calls or in the DB when the conversation is over.
• Available entities is the same as above.

Enhanced Scenario Creation and Generative AI Integration​

Dynamic Scenario Building: Users can now create detailed conversational scenarios in the "Converse > Scenarios > Create" section. This feature supports intuitive creation and management of dialogue sequences, enabling users to define and refine conversational flows effectively. Generative AI Preview: Leverage cutting-edge Generative AI technology to automatically generate dialogue previews based on specified use cases. This tool helps visualize and optimize the conversation flow before finalizing the scenario, ensuring a more effective user interaction. Real-time Editing and Simulation: The interface provides real-time editing capabilities along with a simulation environment that shows how the AI will interact in various conversation scenarios. Users can immediately test and tweak the generated dialogues, enhancing the development process and ensuring higher accuracy and relevance in real-world applications.