Multi-Channel n8n Chatbot Automation
€250-750 EUR
Posted 1 minute ago
•
Ends in 6 days
€250-750 EUR
Paid on delivery
I want a single n8n workflow that lets customers chat with an AI assistant on our website widget, WhatsApp, and Facebook Messenger. The bot must: • Draw answers from three sources I will provide: – a curated FAQ list (CSV) – live content from selected website pages / PDFs – our order & customer management API to pull status details • Decide in real time whether to – answer common questions with an LLM (OpenAI or similar) – check order or account status and return it to the user – hand complex or sensitive issues to a human agent, passing the full chat transcript to Zendesk (or a webhook you suggest). Key expectations – All logic lives inside n8n; no hard-coding in external scripts. – Clean, well-named nodes and sub-workflows so my team can maintain them. – Channel connectors configured and tested with my sandbox credentials. – Clear documentation: an exported .json workflow file, environment variable list, and a short step-by-step guide on how to deploy and modify knowledge sources. I will supply API keys, sample data, and wording. The job is done when I can import your workflow, swap in production keys, and watch the assistant handle each scenario end-to-end.
Project ID: 40826002
About the project
64 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
64 freelancers are bidding on average €359 EUR for this job


Hi, I am Elias from the sunny state of Florida. I understand you're looking to create a single n8n workflow for an AI assistant on your website. The goal is to streamline customer interactions through automation while ensuring a seamless user experience. To achieve this, I propose leveraging n8n's powerful integrations to connect your website widget with the AI backend effectively. By using PHP and JavaScript, we can manage user inputs and responses efficiently. Python can be employed for any custom AI logic needed, ensuring the chatbot operates intelligently and responds appropriately to user queries. Key challenges will include ensuring smooth API integration and maintaining performance during high traffic. I will implement robust error handling and logging to monitor interactions, which will help in future troubleshooting and performance optimization. A few questions to clarify the scope: Q1 – What specific functionalities do you envision for the AI assistant? Q2 – Are there particular APIs or services you would like to integrate? Q3 – How do you plan to handle data privacy and user permissions? I'm excited about the potential of this project and look forward to collaborating with you. Looking forward to hearing from you.


Hi, you want one assistant that answers customers the same way on your site widget and on two messaging channels, looks up real order status, and passes the hard cases to a person with the full conversation. The part that decides whether it works is the hand-over rule: what counts as "complex or sensitive" must be written down as explicit conditions, not left to the model's mood. How I would build it, all inside n8n: 1. One router sub-workflow. Each channel trigger normalises the message into one shape (channel, user, text), so the logic exists once, not three times. 2. Knowledge: your FAQ file and the selected pages and PDFs go into a vector store that n8n refreshes on a schedule; answers must come from those passages, otherwise the bot says it does not know and offers a person. 3. Order lookups through your order API with read-only access, returning only the fields the customer may see, after a simple identity check such as order number plus the email used. 4. Hand-over: sensitive topics (payments, complaints, legal, anger) and repeated failed answers create a Zendesk ticket with the full transcript. 5. Clean, named nodes, an exported JSON, an environment variable list and a short guide to swap production keys and add knowledge sources. I test each channel with your sandbox credentials. I work as a single senior point of contact, and my Freelancer account has a 4.98 rating from 51 reviews. I can share concrete examples in the Freelancer chat. Price 610 EUR in 14 days. Messaging channel approval by the providers is outside my control and I would plan for it. Question: which two messaging channels are live today, and does your order API already support lookup by order number?


Your business needs a unified AI chatbot that handles customer conversations across your website, WhatsApp, and Facebook Messenger while keeping all automation logic inside n8n. I’d build a modular workflow using OpenAI, your FAQ CSV, website/PDF knowledge sources, and order management API, with intelligent routing for common questions, order status requests, and human escalation. **Deliverables:** Complete n8n workflow, website/WhatsApp/Messenger integrations, FAQ and document-based knowledge retrieval, LLM-powered responses, live order and account status checks, Zendesk/webhook escalation with full chat transcripts, clean reusable sub-workflows, sandbox testing, exported JSON workflow, environment variable list, and deployment and maintenance documentation. Ready to start ASAP.


Hi there, To give customers consistent answers across your website widget, WhatsApp, and Facebook Messenger, I can build one maintainable n8n workflow that routes each message to the right source or a human agent. I’ll normalize incoming messages, use your FAQ and selected pages/PDFs for grounded answers, and call your order API only when a status request requires it. The assistant will avoid inventing order details, verify the customer before sharing account information, and escalate sensitive or uncertain cases with the full transcript sent to Zendesk or your preferred webhook. I’ll keep the logic in clearly named n8n nodes and sub-workflows, then test each channel and decision path using your sandbox credentials. You’ll receive the importable workflow JSON, required environment variables, and a concise guide for switching keys and updating knowledge sources. Could you share your widget’s webhook format and the API’s customer-verification method? Those will determine the safest connector setup.


Hello, {{{ I HAVE CREATED SIMILAR APPS BEFORE AND I CAN SHOW YOU }}}} I’ve reviewed your requirements and understand the need for a unified n8n AI chatbot connecting your website widget, WhatsApp and Facebook Messenger. I have 10+ years of experience in automation, AI integrations, n8n, OpenAI APIs and backend development. I will build the complete workflow with FAQ CSV, website/PDF knowledge retrieval, live order and customer API lookups, intelligent query routing and human escalation through Zendesk or webhooks. The workflow will include conversation context, secure API handling, full transcript transfer, error handling and clearly organized reusable sub-workflows. I will configure and test all three channels using your sandbox credentials. I eagerly await your response to discuss the APIs and start with the workflow architecture. Thanks & regards, Karuna


Hi there, Building a multi-channel AI chatbot that seamlessly switches between static knowledge bases, dynamic API lookups, and human hand-offs without external custom scripts can quickly lead to messy, unmaintainable node webs if not architected with clean sub-workflows from the start. I will build a robust, production-ready n8n workflow that unifies your website widget, WhatsApp, and Facebook Messenger into a single intelligent routing engine, cleanly parsing your CSV, live web/PDF data, and order API while integrating smoothly with Zendesk for escalations. Here are my questions: 1. Which specific website widget and live chat interface do you currently use for the web channel? 2. Do you already have a preferred vector store set up within n8n for the website pages and PDFs, or would you like me to configure one using an integrated node? Drop me a message so we can get started right away.


Hello, After reading through your project details, I grasp your goal of building a unified n8n AI assistant across website chat, WhatsApp, and Facebook Messenger that answers from your CSV FAQ, web/PDF sources, and order/customer API, dynamically routes between LLM replies, live account lookups, and Zendesk transcript handoffs, and includes clean sub-workflows, environment variables, a .json export, and a setup guide. This is right up my alley. Over the years, I have engineered multi-channel bots, API integrations, and AI routing workflows utilizing n8n, OpenAI API, WhatsApp Cloud API, Meta Messenger, REST/webhook nodes, and Zendesk APIs. My strategy relies on normalizing incoming messages into a unified schema, using a structured router node to separate FAQ/doc retrieval, order/account API lookups, and Zendesk escalation with full transcripts, keeping all logic inside named n8n nodes and sub-workflows, and testing end-to-end against your sandbox keys. To tailor my plan, please share: 1️⃣: Which WhatsApp provider (such as Meta Cloud API, Twilio, or 360dialog) and chat widget are you using, and does your order API require token refresh or custom headers? 2️⃣: How frequently does your CSV FAQ or web/PDF knowledge base update, and should Zendesk create a new ticket or append to an existing open ticket? Message me in the chat to confirm your sandbox details and get started. Best Regards, Rabia Shaikh.


Hi there, I'd like to clarify the requirements for implementing the n8n workflow to enable multi-channel chatbot automation across your website, WhatsApp, and Facebook Messenger. Given your need for real-time decision-making, I would leverage n8n's capabilities to route queries effectively based on whether they draw from your curated FAQ, live site content, or require API integration for order status checks. I plan to ensure that all logic remains within n8n using clean, well-organized nodes, enabling your team to maintain the workflow easily. Additionally, I'll deliver comprehensive documentation, including an exported .json workflow file and a step-by-step guide for deployment. Looking forward to your response. Regards, Ali


Hi, I've built multi-channel chatbot workflows in n8n before, so I understand the routing logic and how to keep everything maintainable inside the platform without external scripts. I'll design a single workflow that ingests your FAQ CSV, pulls live content via web scraping or PDF parsing, queries your customer API for order status, and uses conditional logic to route messages—either to an LLM for standard questions, directly to your API for account lookups, or to Zendesk when a human handoff is needed. Each decision point will be clearly named so your team can adjust thresholds and knowledge sources later without touching code. I'll configure and test the connectors for your website widget, WhatsApp, and Facebook Messenger using sandbox credentials you provide, then validate the full flow end-to-end with real chat scenarios before handoff. You'll get the exported workflow file, a documented list of environment variables, and a deployment guide that shows exactly where to swap in production keys and how to update your FAQ or content sources. I'm available now and will stay on for any adjustments once you've tested it live.


Routing logic is where multi-channel bots usually fall apart, so I'd build one intent router sub-workflow that all three channels feed into, instead of duplicating logic per channel. I'll wire the website widget, WhatsApp and Messenger into n8n, with a RAG node pulling from your FAQ CSV, pages and PDFs, an order-status branch hitting your API, and a Zendesk handoff passing the full transcript. Vector store for the PDFs beats raw prompt stuffing once content grows. 1) Which vector store do you prefer, Supabase, Qdrant or Pinecone? 2) Is the order API REST with a token, or something else? Message me and we can lock the plan down. Shayan


Bonjour, Fernando du Mexique ici. L'enjeu de ce projet est de faire fonctionner un seul assistant intelligent sur trois canaux, sans perdre le contexte des conversations ni donner de mauvaises informations sur les commandes des clients. Je vous propose de construire le workflow directement dans n8n, avec des sous-workflows clairement organisés pour le widget web, WhatsApp et Facebook Messenger. L'assistant pourra exploiter votre FAQ CSV, les pages web et PDF, ainsi que votre API de commandes. Je mettrai en place une logique de routage pour distinguer les questions courantes, les demandes nécessitant une vérification de commande et les situations qui doivent être transmises à un agent humain avec l'historique complet. Je porterai une attention particulière à la gestion des erreurs, aux réponses fiables et à la facilité de maintenance. Vous recevrez le workflow JSON exportable, la liste des variables d'environnement et un guide de déploiement. Je serais ravi de discuter de vos API et de vos environnements de test. Fernando


The key to this project is giving customers consistent answers across all three channels while ensuring order information is retrieved securely and sensitive issues reach a human agent. I have hands-on experience building complex n8n and Make automations involving Zoho, Shopify, Airtable, and Gmail, including API integrations and webhook-based workflows. I'll build a modular n8n architecture connecting website chat, WhatsApp, and Facebook Messenger. The assistant will use your FAQ CSV and indexed website/PDF content for knowledge-based answers, query your order API when customer-specific information is needed, and escalate complex cases to Zendesk with the full conversation transcript. I'll also implement channel identification, conversation context, error handling, secure API authentication, and clearly named nodes and sub-workflows. You'll receive importable n8n JSON, environment variable documentation, and a deployment guide covering knowledge updates and production configuration. Do you already have WhatsApp Business API and Facebook Messenger credentials configured?


Hello, Routing omnichannel messages from web, WhatsApp, and Messenger into a single n8n workflow requires a central message router that normalizes incoming payloads before hitting the intent classification node. Using OpenAI function calling or structured output parsers will allow the agent to decide dynamically whether to query your FAQ vector store, scrape live site pages, or fetch customer data from your order management API. If intent confidence falls below a set threshold or the user asks for a human, the workflow can trigger the Zendesk API to create a ticket and attach the complete execution history. Maintaining clean sub-workflows for each channel and data source will keep the main canvas maintainable for your team. How are you planning to handle session state persistence across WhatsApp and the website widget for returning customers? Have a nice day


Hello, Your setup really needs one shared brain behind three doors: the website widget, WhatsApp, and Messenger. I would build that as a single n8n parent flow with small, clearly named sub-workflows for each channel and each knowledge source, so your team can change one piece without the whole thing falling over. On the knowledge side I would load your FAQ CSV into a searchable store, pull live text from the pages and PDFs you choose, and wire your order API as a tool the bot can call when someone asks about status. A router step then decides in real time: common questions go to the LLM, order or account questions hit the API, and anything complex or sensitive gets handed to Zendesk (or a webhook you prefer) with the full chat transcript attached. All logic stays inside n8n, no outside hard-coded scripts. I will configure and test the channel connectors with your sandbox credentials, then hand over the exported JSON, an environment variable list, and a short step-by-step guide so you can swap production keys and walk through each scenario end to end. If that sounds like the path you want, I am ready to start as soon as you share the sample data and keys.


Hello, I will create a comprehensive n8n workflow that integrates an AI chatbot across your website widget, WhatsApp, and Facebook Messenger. This solution will utilize your provided FAQ list, live website content, and API to ensure accurate responses for common inquiries and order status. To achieve this, I recommend using n8n for workflow automation, along with API integration to connect your order management system and ensure real-time data access. My experience with n8n and API setups will facilitate a smooth implementation. I will develop and thoroughly test the workflow for all scenarios, ensuring seamless integration with your channels. The process will include clear documentation and well-structured nodes for easy maintenance. I can deliver this within 3 days for the budget of 250 EUR. Are the API endpoints and data formats confirmed for smooth integration? Please feel free to reach out to me. I look forward to working with you. Best regards, Billy Bryan


⭐⭐⭐⭐⭐ Hi there, Could you confirm that you need one maintainable workflow coordinating website chat, WhatsApp, Messenger, knowledge sources, order APIs, and human escalation? The key priorities are reliable intent routing, accurate retrieval from FAQs and live pages/PDFs, secure order lookups, and complete Zendesk handoff transcripts. Each scenario should be testable with your sandbox credentials before production deployment. My AI product work includes OpenAI and Python-based integrations, API-connected workflows, and maintainable SaaS automation. I can learn and configure n8n for this implementation without relying on external scripts. ❓ Which n8n hosting setup and version will we use? ❓ Do your order API and Zendesk webhook already provide sandbox documentation? The technical focus is connecting each channel, knowledge source, LLM, order API, and escalation path with clear reusable steps. I’m ready to begin immediately and will provide the workflow export, variables, and deployment guide. I look forward to your reply and discussing the implementation. Best regards, Phuong Nam D


Hello, I’d love to build this for you. I’ve recently worked on a similar n8n setup where a single workflow handled incoming messages from web chat and WhatsApp, pulled answers from an FAQ sheet plus live website content, and passed order-related requests to an API before escalating sensitive chats to Zendesk with the full transcript. That project needed clean node structure, modular sub-workflows, and easy handover for the client team, so I’m very comfortable with the kind of maintainable build you’re after. For your bot, I’d keep the logic fully inside n8n, with clear routing for FAQ lookup, knowledge retrieval, API status checks, and human handoff. I’d also make sure the workflow is documented properly so your team can update sources without breaking anything. A couple of quick questions: 1) Do you already have a preferred Zendesk handoff format, or should I suggest the cleanest webhook structure? 2) For the knowledge source, do you want website/PDF content indexed inside n8n or fetched live on each request? If helpful, I’d be glad to jump on a call and discuss the flow. Best regards, Mehreen


I can build a maintainable, import-ready n8n workflow that unifies your website widget, WhatsApp, and Facebook Messenger into one AI support experience. The workflow will route each conversation intelligently: answer FAQ questions using your curated CSV, retrieve current information from selected webpages and PDFs, query your order/customer API for secure status updates, and escalate sensitive or complex cases with the complete transcript sent to Zendesk or a suitable webhook. I’ll keep the implementation entirely inside n8n, using clearly named nodes, reusable sub-workflows, structured error handling, and channel-specific testing with your sandbox credentials. You’ll receive the exported JSON workflow, environment-variable checklist, deployment guide, and instructions for updating knowledge sources without rebuilding the automation. The final handoff will be tested against each scenario so you can replace sandbox keys with production credentials and operate confidently from day one. Happy to turn this into a clean, maintainable support workflow.


Hello, I understand you need a single n8n workflow that connects your website widget, WhatsApp, and Facebook Messenger to an AI assistant capable of pulling answers from a CSV, live pages/PDFs, and your order API—then routing between LLM answers, order lookups, or human handoff to Zendesk. The main challenge is keeping all logic inside n8n while ensuring clean, maintainable sub-workflows and reliable real-time decision routing across three channels. I have a few quick questions: - Are you using n8n Cloud or self-hosted? - Which website widget do you plan to use (e.g., custom embed, Chatwoot, etc.)? - Do you already have a preferred LLM provider, or should I default to OpenAI? I have built similar multi-channel automation bots using n8n, API integrations, and LLM fallback logic. I can deliver the exported workflow JSON, documentation, and a tested sandbox setup you can swap to production quickly. Best regards, Carlos.


Hi there, I saw your project for a multi-channel n8n chatbot and it sounds like a great fit. I understand you need a single workflow that can pull information from a CSV FAQ, website content, and your order API, then decide whether to answer with an LLM, check order status, or hand off to a human agent via Zendesk. My first step would be to set up a basic n8n structure to test the API integrations you'll provide, ensuring data can be accessed correctly. I work out of Lagos, Nigeria, and have experience with PHP, Python, and JavaScript, which can be helpful for any API work. Happy to chat more about how I can build this for you.
内容来自 Freelancer News,版权归原作者所有。
阅读原文 ↗