n8n AI Chat Support Automation
$250-750 USD
Posted 1 minute ago
•
Ends in 6 days
$250-750 USD
Paid on delivery
I need a complete n8n workflow that plugs a website-chat widget into two large-language-model agents—Claude and Gemini—so customers can be guided through choosing or using our product without waiting for a human. The flow must: • Receive each incoming chat message, detect intent, and decide whether Claude or Gemini will produce the reply (I will share model keys). • Keep context across the conversation inside n8n so follow-up questions reference previous answers. • Draw on our existing help centre articles and product data that I’ll provide in Markdown/CSV. • Hand off seamlessly to a live agent whenever the user types a keyword such as “human” or the confidence score drops below a threshold we set together. Acceptance criteria • Deployed n8n workflow file (.json) importable on my self-hosted instance. • Connection instructions and environment variables clearly documented. • A short Loom or text walkthrough showing a test chat where the bot helps a customer pick the right plan, then follows up with usage tips. • All code, prompts and decision logic delivered in a Git repo so I can iterate later. If you have experience chaining multiple models inside n8n or similar orchestration tools, I’d love to see it.
Project ID: 40823077
About the project
50 proposals
Open for bidding
Remote project
Active 4 secs ago
Place your bid
Bid amount Email addressBenefits of bidding on Freelancer
Set your budget and timeframe
It's free to sign up and bid on jobs
50 freelancers are bidding on average $363 USD for this job


AI automation and n8n are core services at 360 Elevate, supported by an in-house team of 60+ professionals across development, QA, and automation. This makes me well suited to build your website-chat workflow with model routing, persistent conversation context, knowledge from Markdown/CSV content, and escalation to a human agent. My n8n, JavaScript, PHP, and AI Chatbot Development skills align directly with the required implementation. I can deliver an importable n8n JSON workflow, organized prompts and decision logic in a Git repository, plus clear environment-variable and connection documentation. The workflow can incorporate Claude and Gemini routing, keyword-based handoff, and confidence-threshold escalation while keeping the conversation history available for plan recommendations and follow-up usage guidance. Choose me for a structured, maintainable automation backed by broad experience in AI systems, software development, and cross-platform business automation. I’m ready to deliver the workflow and walkthrough in a form your team can continue iterating on.


Hi, I am Elias from the sunny state of Florida. I understand you're looking to create a complete n8n workflow that integrates a website chat widget with two large language model agents. This is a great way to enhance customer interaction and streamline support processes. My approach would involve first mapping the workflow to ensure seamless communication between the chat widget and the LLMs. I’d leverage n8n's capabilities to manage data flow, using JSON for structured data exchanges between components. Ensuring robust error handling and monitoring will be crucial for reliability. In terms of technologies, I would utilize PHP and JavaScript to handle backend processes and frontend interactions efficiently. Moreover, setting up proper API integrations will allow the LLMs to respond accurately and contextually based on user queries. A few questions to clarify the scope: Q1 – What specific features do you want in the chat widget, such as user authentication or analytics? Q2 – Are there particular LLMs you plan to use, or are you open to recommendations? Q3 – What is the desired timeline for implementing this workflow? Looking forward to hearing from you.


Hi, Routing traffic between Claude and Gemini based on real-time intent detection requires a custom JavaScript node right after the webhook to evaluate query complexity before hitting the LLM nodes. I will build this exact architecture in n8n, using a Redis or built-in memory node to preserve chat context across turns and ingest your Markdown and CSV help centre data into a vector store for accurate retrieval. Chaining multiple language models with conditional logic and fallback triggers is work I handle daily, and my plan is to set up structured prompt templates, a confidence score filter, and a keyword intercept for human handoff. I will make sure not to drop context during the model switch so the conversation stays natural. For $425, delivery takes 7 days and includes the importable n8n workflow file, environment documentation, the Git repository with all prompts and decision logic, and a testing walkthrough. Tell me which vector database you prefer to use for your help centre articles.


Hi, I will build an n8n chat-support workflow that connects your website widget to Claude and Gemini, routes each message through intent detection, and selects the best model for the reply. I’ll store conversation state in n8n for follow-ups, ingest your Markdown/CSV knowledge base, and add a live-agent handoff when the user asks for a human or confidence drops. I’ll deliver the importable .json workflow, setup docs with required environment variables, prompts/decision logic in a Git repo, and a walkthrough showing plan selection plus usage guidance. Please start the chat so we can discuss the exact handoff rules and knowledge-base structure. Regards, Sai Bhaskar


My 12+ years delivering business automation and AI chatbot solutions make me well suited to build your n8n-based support system. I can connect the website chat widget with Claude and Gemini, use n8n to route messages by intent, retain conversation context, and create a smooth path to human support when a keyword or confidence threshold requires escalation. My experience includes AI-powered and hybrid chatbot setup, chatbot integrations with CRM and business systems, and full workflow automation. Combined with JavaScript, n8n, PHP and XML skills, I can deliver an importable workflow that is practical to maintain and extend. I will also provide the connection and environment-variable documentation, prompts and decision logic in a Git repository, plus a concise walkthrough demonstrating plan selection and follow-up usage guidance. I am ready to build a reliable, well-documented multi-model support workflow that your team can continue iterating on. Choose me to turn your product data and help-centre content into a useful automated customer experience.


Hi there, You need an automated chat solution using n8n to connect customers with the Claude and Gemini language models. This setup will guide users through product selection and questions, providing context-aware conversations. I have hands-on experience creating complex n8n workflows for chat automation. How I would approach it: - Develop a workflow to receive incoming chat messages and route them to either Claude or Gemini based on intent detection. - Implement conversation context management within n8n to ensure relevant follow-up responses. - Integrate the provided help center articles and product data for contextual support. - Create a seamless transition to live agents when specific keywords are detected or confidence thresholds are met. I have over three years of experience with n8n and workflow automation. I document every project, providing clear instructions for future use. Which keywords should trigger the handoff to a live agent? I look forward to delivering an efficient automation solution that meets your needs.


Hi, I understand you need a complete chat automation flow that can route questions between Claude and Gemini, keep conversation context, use your Markdown/CSV product information, and pass the chat to a human when needed. I will organize the decision logic, prompts, confidence checks, and handoff rules clearly so the workflow is easy to update. My work with JavaScript, Java, PHP, and Automation will support the integration logic, data handling, testing, and Git delivery. I will also provide the importable workflow, setup notes, environment variable details, and a clear test walkthrough. n8n-specific node and deployment details should be confirmed before implementation. Which website chat widget and live-agent platform should the n8n workflow connect to, and does your self-hosted n8n allow custom nodes or only built-in nodes? Please start the chat and let’s discuss the widget, data format, and handoff setup.


Hi there, You need a complete n8n workflow to integrate a chat widget with the Claude and Gemini AI agents for your product inquiries. This requirement aligns well with my expertise in developing AI chatbot solutions and workflow automation, ensuring users receive timely and accurate assistance without human intervention. How I would approach it: - Set up n8n to receive and parse incoming chat messages, utilizing intent detection to choose between Claude and Gemini. - Maintain conversational context using n8n's features to ensure follow-up inquiries are coherent. - Integrate the provided help center articles and product data for enriched responses. - Establish a smooth handoff protocol to a live agent based on user keywords or confidence scores. I deliver well-documented workflows and a Git repository for future iterations. How do you prefer to approach the integration of your help center content?


Hi there, I'm excited about your project to create a robust n8n workflow integrating website chat with both Claude and Gemini LLMs. I understand the goal is to automate customer support by intelligently routing queries, maintaining conversation context, and leveraging your help centre content—all while ensuring a smooth handoff to live agents when needed. With extensive experience in n8n workflow automation, AI chatbot development, and integrating multiple LLMs, I can design a solution that dynamically detects user intent and assigns the optimal model for each response. My approach will use n8n’s native nodes, custom JavaScript, and HTTP requests to handle context management, intent routing, and fallback triggers based on keywords or confidence scores. I’ll securely connect your model APIs using environment variables, and process your Markdown/CSV content for accurate, context-aware responses. Comprehensive documentation and a Loom walkthrough will make deployment and future iterations seamless for you. All logic, code, and prompts will be version-controlled in Git for easy collaboration. If you’d like, I can share examples of past multi-model orchestration projects where I implemented similar decision flows and live agent escalations. I’m confident I can deliver a flexible, well-documented workflow that meets your criteria and sets a strong foundation for future enhancements. Looking forward to collaborating! Best Regards


Fernando from Mexico here. The important part of this project is making Claude and Gemini work as one reliable support system, rather than simply connecting two APIs and switching between them. I'd build the workflow in n8n with an intent-routing stage, model-specific agents, and shared conversation memory so follow-up questions retain context. Your Markdown and CSV resources would be organized into a searchable knowledge base, allowing answers to reflect your actual products and help articles rather than relying only on general model knowledge. I'd also implement confidence-based escalation and an explicit human handoff, ensuring the bot stops responding when a live agent takes over. Error handling, execution logs, configurable prompts, and secure API credentials would be included. I'll deliver the importable n8n JSON, Git repository, setup documentation, and a tested walkthrough covering plan selection and follow-up support. One detail I'd like to clarify first is how your website chat currently connects to the backend. Fernando


Routing website chat traffic through an advanced n8n workflow while balancing inputs across two different large language models requires precise JSON parsing and context management. We specialize in building custom automation workflows, including OpenAI-based chatbots and n8n integrations that connect seamlessly with web frontends. Our approach: 1. Map the chat widget payload structure in JavaScript and JSON. 2. Build the core n8n workflow with conditional routing to both large language model agents. 3. Configure response aggregation and error handling. 4. Test live chat interactions and refine performance. Are you using a specific hosting platform for your current chat widget endpoint? Ahsan Ali Khan


Hello, I will create an n8n workflow to integrate your website chat widget with Claude and Gemini, enabling automated customer support by detecting intent and managing conversation context. This will ensure users receive immediate assistance without waiting for a human agent. Given my experience with PHP, JavaScript, and n8n, I recommend using n8n's built-in nodes to handle incoming messages and API calls to both large-language models. This will allow for efficient intent detection and seamless context management throughout the conversation. I will develop the workflow, ensuring it meets all acceptance criteria, including the deployment of the .json file and comprehensive documentation. I plan to test the workflow by simulating customer interactions, demonstrating its capabilities in a Loom video. Please confirm any specific keywords you'd like for agent handoff and if you have a preferred method for accessing help center articles. Please feel free to reach out to me. I look forward to working with you. Best regards, Billy Bryan


Hello, I’d be glad to help you build this n8n chat support automation. I’ve worked on similar AI support workflows where incoming website messages are routed through different LLMs based on intent, with conversation memory preserved so the assistant can answer naturally across follow-ups. I’ve also built knowledge-driven automations that pull from help docs and product data, then escalate to a human when the conversation needs a handoff. For example, I recently set up a support flow where one model handled product selection questions and another handled usage guidance, while the workflow stored context inside the automation layer and switched to live support when the user asked for a person or when confidence was low. The result was a much smoother customer experience and fewer repetitive replies. A couple of things I’d like to confirm: 1. Do you already have a preferred structure for the Markdown/CSV knowledge base, or should I design the ingestion format? 2. For handoff, should the live agent route into an existing inbox/helpdesk, or just trigger a notification for now? If helpful, I can also suggest the best way to structure the confidence thresholds and prompt routing logic. Let’s connect for a quick call to discuss the flow. Best regards, Mehreen


⚠️ If you're not happy, you don’t pay. ⚠️ Hi Client, thank you for checking my proposal and sharing the detailed project brief. I can build your n8n workflow using JavaScript and REST APIs with a scalable and user-friendly design. I will deliver: • Intent detection for incoming messages using Claude and Gemini • Context management to ensure conversation continuity in n8n • Integration of your help center articles and product data from Markdown/CSV • Automatic handoff to a live agent based on keywords or confidence scores • A .json workflow file for easy deployment on your self-hosted instance • Clear documentation for connection instructions and environment variables • A Loom walkthrough demonstrating the workflow in action You will also receive: • A Git repository with all code, prompts, and decision logic for future iterations I am confident I can execute your vision professionally and efficiently. Looking forward to discussing the timeline and next steps. Best regards, Chirag Pipal


HELLO, "n8n AI Chat Support Automation" - you need Claude and Gemini working together with shared conversation context, product knowledge, and reliable human handoff. I'll build the complete n8n workflow to receive chat messages, route requests between both models, retain conversation history, and retrieve relevant answers from your Markdown and CSV knowledge base. I'll configure confidence-based escalation and keyword detection so customers reach a human whenever the bot cannot answer reliably. I'll also account for missing product information and API failures to prevent fabricated answers or broken conversations. You'll receive the importable JSON workflow, documented environment variables, Git repository, and a tested walkthrough. Should the website chat widget already provide a webhook-compatible integration, or does that connection need to be built as well? I appreciate your consideration and look forward to discussing the workflow. THANKS


Chat routing logic: Routing chat messages between Claude and Gemini on your self-hosted instance risks dropped context if memory nodes are misconfigured. I build this using the n8n Advanced AI node with a persistent Redis chat memory and Markdown embeddings. Over 26 years of industry experience, with 5 stars on every job completed here, including a full-stack MVP built on a complex set of business rules. We fix what done means as a written checklist up front, and you release payment once the output passes it. My work and reviews: https://www.freelancer.com/u/azharjafri Which vector store or database do you plan to use for indexing your CSV and Markdown help articles?


A support chat that switches between Claude and Gemini inside n8n is a build I'm glad to take on for you. I can deliver this end to end on your self-hosted instance, using your own model keys. I'll stay on it until the chat behaves the way you want it to. Which chat widget is your website using at the moment? Happy to talk through the details whenever suits you. Sela


Hi, I’ve reviewed your n8n AI Chat Support Automation and understand what you’re looking to achieve. I can help you handle the required work with a clean, reliable, and professional approach. I have experience with PHP, JavaScript, XML, JSON, AI Chatbot Development, Large Language Models (LLMs), n8n, Workflow Automation and can work effectively with your existing setup, requirements, and workflow. I’ll focus on proper implementation, responsive functionality, attention to detail, and thorough testing to make sure everything works as expected. I’ll also make sure the completed work is aligned with your requirements rather than just providing a quick or temporary solution. I’m available to start immediately and can work according to your required timeline. Best regards, Khadija Tul Kubra


Hello, The key challenge here is not simply connecting two models; it is making routing, conversation memory, help-center retrieval, and human escalation behave predictably in one customer journey. I would structure the first milestone around a tested workflow that classifies intent, selects the appropriate model, preserves context, grounds answers in your Markdown/CSV content, and triggers handoff on keywords or configurable confidence thresholds. My background includes LLM integrations, AI agents, document processing, and automation systems. I can provide the importable workflow, prompts, environment-variable documentation, repository structure, and a concise test walkthrough covering plan selection and follow-up usage guidance. I would target a working first milestone within 7 days, with the implementation organized for later iteration. Best regards, Bohdan


Hola, "automatizacion de soporte con Claude y Gemini" - respuestas contextualizadas, seleccion de productos y derivacion fiable a agentes humanos. Implementare personalmente el flujo completo en n8n: recepcion del chat, deteccion de intencion, enrutamiento entre modelos, memoria conversacional, consulta de articulos Markdown y datos CSV, y transferencia a soporte humano. Tambien contemplare respuestas de baja confianza, errores o tiempos de espera de las API y mensajes duplicados para evitar respuestas inconsistentes. Entregare el JSON importable, variables de entorno, instrucciones de conexion, repositorio Git y una guia de pruebas en texto. La logica de confianza debe combinar umbrales configurables con reglas de seguridad, no depender unicamente de la puntuacion declarada por el modelo. Que plataforma proporciona el widget de chat y como debe recibir el agente humano las conversaciones transferidas? Saludos cordiales, Beate
内容来自 Freelancer News,版权归原作者所有。
阅读原文 ↗