Feature Requests

We’re seeing 60+ “write this article” prompts for topics way outside our domain — many look bot-generated. This clogs our queue, wastes time, and risks diluting the knowledge base. What we need: Smarter controls so AI suggestions only surface for content we actually want to cover Stronger bot protection to block junk searches from triggering new-article prompts A way to exclude certain topics entirely from the “auto-suggest an article” logic Just because someone searches for something and doesn’t find it doesn’t mean we should create an article — even if it gives a short-term SEO bump. Why this matters: Keeps AI focused on suggesting high-value content without polluting KB quality and brand relevance.

Summary & Executive Brief We propose that HelpJuice develops and provides a native Model Context Protocol (MCP) server. This would allow AI agents (like those built on Claude, ChatGPT, etc.) to securely and reliably read data from a customer's HelpJuice knowledge base. In essence, this feature would transform a static KB into a dynamic, AI-ready data source, enabling customers to build powerful, context-aware AI assistants that have direct, sanctioned access to their proprietary documentation. The Problem / Opportunity The AI landscape is rapidly shifting from chat interfaces to AI agents that perform tasks autonomously. These agents need access to contextual, proprietary data to be useful. Currently, providing this data involves: Costly and Complex Integrations: Building and maintaining custom APIs and connectors for each AI platform. Unreliable Methods: Using web scraping, which is fragile and breaks with UI changes. Context Window Limitations: Manually pasting KB articles into prompts is inefficient and doesn't scale. Security Concerns: Granting broad API keys or using unsecured methods to expose data. An MCP server solves this elegantly by providing a standardized, secure, and managed bridge between the AI and the HelpJuice KB. Proposed Solution: Native HelpJuice MCP Server Develop a dedicated MCP server that exposes a controlled set of "tools" for an AI to interact with the knowledge base. The core tools would initially be focused on data retrieval. Key MCP Tools to Expose: search_knowledge_base(query: string, category?: string) Performs a semantic and/or keyword search across the entire KB or a specific category. Returns a list of relevant articles with titles, snippets, and URLs. get_article_by_id(article_id: string) Fetches the full, clean text content of a specific article by its internal ID. list_categories() Allows the AI to understand the structure of the KB by listing all top-level categories. get_articles_by_category(category_id: string) Retrieves all articles within a specific category, useful for comprehensive research. Benefits for HelpJuice Customers Build Smarter AI Assistants: Power internal help desks, customer support bots, and onboarding coaches with accurate, real-time knowledge. Drastically Reduced Development Time: Customers can connect their KB to an AI agent in hours, not weeks, using a standard protocol. Enhanced Data Security: The MCP model provides a controlled gateway. Customers never have to expose full database credentials to the AI. Improved AI Accuracy: Eliminates hallucinations by providing the AI with a direct, authoritative source of truth. Competitive Advantage: Positions HelpJuice as a forward-thinking, AI-native platform, not just a documentation tool. Benefits for HelpJuice Strategic Positioning: Becomes the "knowledge backbone" for the AI ecosystem. Increased Stickiness: Deeper integration into a customer's AI/automation stack makes the product indispensable. New Revenue Streams: This could be a premium feature or part of a higher-tier enterprise plan. Future-Proof Foundation: The MCP framework makes it easy to add more complex tools later (e.g., draft_article, suggest_edit). Technical Implementation Outline MCP Server: A standalone service (e.g., in Node.js or Python) that uses the HelpJuice API for data access. Authentication: Utilizes HelpJuice's existing API key system. The customer provides their HelpJuice subdomain and API key when configuring the MCP connection in their AI platform. Deployment: The server can be provided as: A hosted service by HelpJuice (simplest for customers). A Docker container that customers can self-host for maximum data control. Documentation: Provide clear setup guides for connecting to popular AI platforms like Claude Desktop, Cursor, and others that adopt MCP. Example User Story As a Customer Support Manager, I want our AI support agent to have instant access to our latest HelpJuice documentation, So that it can accurately answer customer questions without hallucination, reducing ticket volume and improving resolution time. Implementation Flow: User asks the AI agent: "What is our refund policy for digital products?" The AI agent uses the configured HelpJuice MCP server's search_knowledge_base tool. The MCP server returns the relevant article(s) from the live KB. The AI agent synthesizes the information and provides a concise, accurate answer to the user, citing the source. Why MCP and Not Just a Better API? While a RESTful API is powerful, MCP is specifically designed for AI agents. It's a standard that is being rapidly adopted by all major AI players. It handles state, resource management, and tool discovery in a way that is native and intuitive for AI models, reducing integration friction significantly.

If we add the Helpbar to the Russian KB, everything will be in Russian except for the following elements: Recommended New Articles Search results for "" x more posts It's not possible to customize it at the moment. The Helpbar UI should follow the KB language.

I would like to request the ability to have multiple chat threads within the AI chatbot. This feature would allow users to: Start and switch between different conversations Recall and continue past conversations Clear or delete specific conversations when no longer needed This would greatly enhance the user experience by making it easier to manage and organize interactions with the bot. Additionally, it ties into a previously requested feature regarding user identification (as referenced in this post https://helpjuice.canny.io/feature-requests/p/record-analytics-for-the-swifty-ai-chabot-and-helpbar ) to help maintain a seamless experience across sessions and chats.

We would like to request an improvement for the AI ​​Chat window. Currently, its size cannot be adjusted, which limits usability. Our proposal: Make the AI ​​Chat window resizable, with an option to open it in full screen. Add support for increasing font size to improve accessibility for users with visual impairments. This feature would significantly enhance the user experience, especially for accessibility needs, and make the chatbot more flexible in different scenarios. Thank you for considering this request!

In the AI Editor Settings section, add the option to disable the related articles suggestions.

Keeping articles accurate as products evolve is tough — and manual reviews don’t always catch outdated info. It would be game-changing to connect Helpjuice with a fact-checking bot (e.g., Swifty or similar) that can: Scan an article and compare it to the latest product data or docs Flag potential discrepancies (e.g., “this setting was renamed in v2.3”) Suggest updates when information looks obsolete or incomplete Provide a simple meter or report that could identify articles that may be outdated, have conflicting information, or are inconsistent with other sources (would need user to specify what "trusted sources" might be) Even if coverage is limited by what the bot can access, automated hints would surface issues long before a customer reports them.

Currently, the same Swifty AI customization is applied to all languages. For example: If we wanted to customize the Welcome Header text in German for the German KB, it will also be applied to the English KB. We should be able to add language-specific settings.

It would be great if we had the ability to adjust these sections from the Swifty Settings menu so that users can adjust it according to Knowledge Base Language.

As a user of Helpjuice, we would like to request the implementation of a feature that allows direct access to the vector database through the Helpjuice API. This capability would enable us to integrate our own software solutions with the data stored in the vector database, enhancing our ability to utilize Helpjuice’s AI functions effectively. Justification: Currently, the Helpjuice API does not provide direct access to the vector database, which limits our ability to fully integrate Helpjuice’s AI capabilities into our custom software solutions. By providing this access, we could leverage the vector data for more advanced data analysis and processing, ultimately improving our workflow and efficiency. Potential Benefits: Enhanced integration possibilities with custom software solutions. Improved data analysis capabilities using the vector database. Greater flexibility in utilizing Helpjuice’s AI features. Increased user satisfaction and productivity. We kindly ask for this feature to be considered for future development. Thank you for your attention to this request. Best regards, Stefan