n8n AI Chat Support Automation
$15-25 USD / hour
Posted 1 minute ago
•
Ends in 6 days
$15-25 USD / hour
I want an end-to-end n8n workflow that captures incoming website-chat and social-media messages, runs them through two cooperative language-model agents—Anthropic Claude and Google Gemini—and sends back accurate, friendly replies to our customers. Here’s the flow I have in mind: messages arrive via the chat widget, Facebook/Instagram DMs, or similar APIs; n8n normalises the payload, then Claude drafts an answer while Gemini refines tone and checks factual accuracy. If the combined confidence score falls below an agreed threshold the workflow must automatically open a support ticket so a human can follow up, and let the customer know their query is queued. I need clean, reusable sub-workflows, clear comments on every node, and environment variables for keys, endpoints, and confidence thresholds. Log each interaction—original message, model responses, confidence metrics—into Postgres (or another SQL layer) so we can audit conversations and retrain later. Deliverables • n8n workflow export (.json) ready for one-click import • README covering setup, env variables, and model-swap instructions • Postgres schema or migration script for conversation logging • Brief video or screenshots showing the automation answering real chat and social messages, plus ticket creation when confidence is low
Project ID: 40821422
About the project
64 proposals
Open for bidding
Remote project
Active 3 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 $20 USD/hour for this job


Hi, I am Elias from the sunny state of Florida. I understand you need an n8n workflow to automate capturing messages from your website chat and social media. This requires a seamless integration of multiple channels and efficient handling of incoming data to ensure prompt responses. To approach this, I would set up n8n to connect with the relevant APIs for your chat and social media platforms, ensuring proper authentication and permissions. The workflow would involve parsing incoming messages, triggering AI responses, and storing conversation data in either MySQL or PostgreSQL for further analysis and reporting. I’d focus on building a reliable system that minimizes latency and maximizes user engagement. A few questions to clarify the scope: Q1 – Which specific social media platforms do you want to integrate? Q2 – What type of AI responses are you aiming for—predefined or dynamic based on user input? Q3 – Do you have any existing infrastructure or tools that we need to integrate with n8n? Looking forward to hearing from you.


Hi, I'm Kausar. You need an end-to-end n8n workflow for automating responses to website chat and social media messages using AI models. I'd start by mapping the entire flow from message capture to response generation and ticket creation for low-confidence queries. I'd create reusable sub-workflows with clear comments for each node, ensuring environment variables are set for keys and thresholds. Logging interactions into Postgres will allow for future audits and training improvements. I'll also provide a README for setup and a video demonstrating the automation in action. What’s the expected response time for customer queries?


Hi, I will build a complete n8n automation that receives website chat and social messages, normalizes each payload, and routes it through two coordinated AI steps: Anthropic Claude for the initial draft and Google Gemini for tone refinement and factual validation. I will add a confidence-scoring branch so low-confidence replies automatically create a support ticket and send the customer a queue notice. I will structure the workflow into reusable sub-workflows, annotate every node with clear comments, and use environment variables for API keys, endpoints, and confidence thresholds. I will also set up Postgres logging for the original message, model outputs, confidence scores, and final action so you can audit and retrain later. You will receive the n8n export, README, SQL schema or migration, and proof of the automation handling real messages. Please start the chat so we can discuss the exact channels, ticketing tool, and logging preferences. Regards, Sai Bhaskar


Greetings, You’re looking to create an n8n workflow that efficiently manages chat and social media messages by integrating two AI models—Claude and Gemini—for accurate, friendly responses. I can help set up a seamless process where incoming messages are normalized, processed for tone and accuracy, and if necessary, escalate to a human for follow-up. My approach would involve building clean, reusable sub-workflows with detailed comments and environment variables for easy management. Capturing interaction logs in Postgres will ensure you can audit and refine the system over time. Deliverables will include an easy-to-import workflow file, a comprehensive README for setup, and a schema for conversation logging, plus visual proof of the automation in action. I’m excited to bring your vision to life! Best regards, Saba Ehsan


Hi there, You need a robust n8n workflow to automate your customer interactions via website chat and social media, integrating AI models for accurate and friendly responses. This project aligns well with my background in AI chatbot development and automation workflows. How I would approach it: - Design the n8n workflow to capture messages from your chat widget and social media platforms, normalising the payloads for processing. - Implement the integration of Anthropic Claude and Google Gemini for drafting and refining responses, including a confidence score mechanism for ticket creation. - Create sub-workflows, ensuring clear comments on each node and defining environment variables for keys and thresholds. - Set up PostgreSQL logging for all interactions, including models' responses and confidence metrics, along with the appropriate schema. What is your preferred confidence threshold for opening support tickets? I can ensure a seamless and effective automation that not only meets your initial needs but also provides a scalable solution for future interactions.


Hi, I understand you need an end-to-end message automation flow that receives customer questions, gets two model checks, sends a clear reply, and routes low-confidence cases to a human. I can help structure the automation logic with Python, PHP, and MySQL, including payload handling, confidence rules, SQL conversation logs, reusable steps, environment settings, and setup notes. Since n8n, Claude, Gemini, and Postgres are not listed in my current tools, I would first confirm the exact access and integration setup before building the final importable workflow. I will keep the flow clean and easy to test, with clear fallback handling for support tickets. Which chat, Facebook/Instagram API, ticketing system, and n8n version should the workflow support first? Please start a chat and let’s discuss the APIs, ticket system, and first test case.


Hi there, I noticed you're focused on creating an n8n workflow that captures incoming messages and utilizes both the Anthropic Claude and Google Gemini models for response generation. I can help you develop a streamlined workflow to handle message normalization and ensure responses are friendly and accurate. I'll include clean, reusable sub-workflows and clear comments for maintainability. Additionally, I can set up the PostgreSQL schema for conversation logging and provide detailed documentation for easy setup, including environment variable management and model substitution instructions. I’d be ready to start on this right away. Looking forward to your response. Regards, Ali


Building AI-powered support systems with both Claude and Gemini integrations is a core Web Crest capability, making this n8n automation a strong match for our team. We specialise in AI chatbots, AI agents, workflow automation, and API integrations, with Python and PHP expertise for reliable custom logic around message processing, confidence handling, and ticket escalation. Our experience with PostgreSQL, REST APIs, cloud technologies, and SaaS development aligns directly with your requirements for normalised chat and social payloads, auditable conversation logs, environment-based configuration, reusable workflows, and maintainable documentation. We can deliver an import-ready n8n JSON workflow, SQL schema or migration, README with model-swapping guidance, and a practical demonstration covering both successful replies and low-confidence human escalation. I would be pleased to build this as a clean, scalable support automation that keeps customers informed while giving your team full visibility into every AI decision.


[DUAL-AGENT N8N AUTOMATION] Hello, The important part of this workflow is not simply connecting Claude and Gemini. It's making sure both agents work together reliably, disagreements are handled properly, and uncertain answers never reach customers as if they were verified. I'm a full-stack developer experienced in Python, PHP, SQL and AI automation. I'd structure this in reusable n8n sub-workflows, with Claude handling the initial response, Gemini reviewing factual accuracy and tone, and a defined decision layer controlling whether to reply or escalate to human support. I'd also implement structured Postgres logging, configurable confidence thresholds, secure environment variables and clear node-level documentation. One detail I'd pay particular attention to is confidence scoring: model-reported confidence alone isn't a reliable measure of factual accuracy, so I'd recommend combining validation rules and agreement checks before allowing automatic replies. You'll receive the importable workflow, database migration, setup documentation and a working demonstration. My proposed hourly rate is $22. Best regards, Fernando


Hello, I would build this as a modular n8n setup where web chat, Facebook, and Instagram all hit one shared normaliser first, so every message looks the same before it reaches the AI layer. Claude drafts the reply, Gemini checks tone and facts, and a clear confidence rule decides whether we send the answer or open a support ticket and tell the customer they are in the queue. Reusable sub-workflows for intake, the dual-model pass, ticket creation, and logging keep the build tidy. Every node gets a short comment, and API keys, endpoints, and the confidence threshold live in environment variables so you can swap models or tighten the bar without rewriting the flow. Each turn (original message, both model outputs, scores) lands in Postgres with a clean schema for audits and later retraining. You get the importable workflow JSON, a README with setup and model-swap notes, the migration script, and a short walkthrough of a real chat plus a low-confidence ticket. I can work 40 hours each week on this. Happy to agree the confidence threshold and channel list with you before we lock the nodes.


Hello, I will develop a robust n8n workflow to seamlessly capture incoming chat and social media messages, integrating Anthropic Claude and Google Gemini for accurate, friendly responses. This will include normalizing payloads, drafting responses, and ensuring fallback support when confidence scores are low. For this project, I recommend utilizing n8n for workflow automation, along with PostgreSQL for logging interactions. My experience with PHP and Python will help in efficiently integrating the language models and creating reusable sub-workflows tailored to your needs. I plan to deliver the n8n workflow export, a detailed README, and a migration script for conversation logging. Testing will involve real messages, ensuring confidence checks trigger support tickets as needed. Please confirm the specific endpoints and keys you will provide for the models. Please feel free to reach out to me. I look forward to working with you. Best regards, Billy Bryan


Hello, I’d love to build this n8n AI chat support automation for you. I’ve worked on similar automation systems where incoming messages from website chat and social channels were routed through AI steps, scored for confidence, and then either answered automatically or pushed to a human ticket queue when a manual review was safer. For this project, I can create a clean, reusable n8n workflow with separate sub-workflows, proper comments on every node, and environment-based configuration for API keys, thresholds, and endpoints. I can also set up the logging layer in PostgreSQL so every conversation, model output, and confidence score is stored neatly for auditing and future retraining. A recent similar build I completed involved an AI support flow using webhook intake, multi-model response checking, and SQL-based conversation logging with fallback escalation to human support. It worked smoothly across different message sources and was designed to be easy to maintain. A couple of quick questions: 1) Do you already have the chat widget and social APIs connected, or should I include the intake setup too? 2) Do you want the ticket handoff to integrate with a specific tool like Jira, Zendesk, or email? I also work extensively with PHP, AI Automation, MySQL, PostgreSQL, and Python, so I can keep the setup practical and production-ready. Could we connect for a quick call to align on the flow? Best regards, Istkhar


I can build the complete n8n support automation, connecting your website chat and social messaging APIs with coordinated Claude and Gemini processing. The workflow will normalize incoming payloads, generate an initial response with Claude, have Gemini refine tone and validate factual consistency, and calculate a configurable confidence result. When confidence falls below your threshold, it will create a support ticket and notify the customer that their request is queued for human follow-up. I’ll structure the solution with reusable sub-workflows, clear node comments, environment-based configuration for credentials and thresholds, and robust error handling. Conversation inputs, model outputs, confidence metrics, routing decisions, and ticket events will be stored in PostgreSQL for auditing and future retraining. Deliverables will include an import-ready n8n JSON export, setup and model-swap documentation, a PostgreSQL schema or migration, and practical proof through screenshots or a short demonstration covering both successful replies and low-confidence escalation.


Hello, After reviewing your requirements, I understand the scope: an end-to-end n8n support workflow with Claude drafting, Gemini reviewing, and human escalation when confidence is low. I have experience with API integrations and AI automation and am available to start right away at $20/hr. One key challenge is avoiding confident but inaccurate replies. I would ground answers in your support content, use structured review scores, and route uncertain cases to a ticket with a customer acknowledgment. Webhook validation, retries, and deduplication would help prevent missed or duplicate replies. I would build reusable, annotated sub-workflows, keep keys and thresholds in environment variables, and log messages, model outputs, and scores in Postgres. Deliverables would include the importable JSON, SQL schema, setup/model-swap README, and a short demo showing replies and escalation. I have a couple of quick questions. • Which chat widget, social channels, and ticketing system should I connect first? • What support documentation should the agents use to verify answers? I would be glad to discuss the details and can start immediately. Best regards, Carlos.


I can set this up in n8n with a Chat Trigger node for website chat, pulling in social messages through their respective APIs, routing through an AI model for responses, and logging everything to MySQL or Postgres for history. Can start today and have a working flow by Thursday. These numbers are based on the post as written, we will refine them after a quick scope conversation. Want me to send a quick scope doc so we can get moving?


Building reliable AI automation with Python, PostgreSQL, and chatbot development is central to my technical skill set, making me well suited to create your end-to-end n8n customer-support workflow. I can connect website chat and social messaging APIs, normalize incoming payloads, coordinate Claude and Gemini responses, and route uncertain conversations to human support. My experience in AI solutions and digital product development supports the reusable, maintainable structure you need. I can deliver clearly commented n8n sub-workflows, environment-based configuration for credentials and thresholds, and PostgreSQL logging for original messages, model outputs, confidence metrics, and ticket events. The README can document setup and model swapping, while the demonstration will show successful replies and low-confidence escalation. TechWagger combines AI automation, chatbot development, Python, and PostgreSQL with a focus on secure, scalable business solutions. I am ready to build this workflow as a practical, auditable support system and would be glad to discuss the project further.


Hi there, You want customers on website chat, Facebook and Instagram to get accurate, friendly replies at any hour, with anything uncertain passed to a human instead of guessed at. I would build this in n8n as separate sub-workflows: one to normalise messages from the chat widget and Meta APIs, one for Claude's draft, one for Gemini's tone and accuracy check, and one for ticketing and logging. Keys, endpoints and confidence thresholds would live in environment variables, and every interaction would be written to Postgres with the original message, both model outputs and the scores. The README would cover setup and how to swap either model later. Which ticketing system should the low-confidence cases go into, and is it already in use by your support team? Send me your helpdesk details and I can show you how the logging schema would look.


HELLO, "n8n AI Chat Support Automation" - you need two AI agents working together to deliver reliable replies and escalate uncertain cases automatically. I'll build the complete n8n workflow connecting website chat and social messaging APIs to Claude and Gemini, with message normalisation, answer refinement, confidence checks, and automatic human-ticket escalation. I'll structure reusable sub-workflows, document every node, and keep API keys, endpoints, and thresholds configurable through environment variables. Postgres will store original messages, model outputs, confidence scores, and interaction history for auditing. I'll also provide the import-ready JSON, setup README, database migration, and a working demonstration. Should low-confidence conversations create tickets in an existing support platform or a new ticketing system? I appreciate your consideration and look forward to discussing the implementation. THANKS


Routing chat and Facebook/Instagram messages through Claude and Gemini, with a confidence gate that escalates weak answers to a human, is a clear and workable design. We would build it as reusable n8n sub-workflows: one for intake and payload normalisation across the chat widget and social APIs, one for the Claude draft and Gemini review step, one for confidence scoring and ticket creation, and one for logging. All keys, endpoints and the confidence threshold will live in environment variables, and every node will carry a comment. The Postgres schema will store the original message, both model responses, the confidence metrics and the ticket status, so you can audit conversations and build a retraining set later. We can start right away and deliver on a weekly schedule with a first milestone covering intake and one model path. We have delivered Salesforce-based support and workflow automation for clients such as NCT Consultants, where manual handling was replaced by automated flows. For this project, we will focus on a working n8n pipeline with clear documentation. Which social channels need to go live first, and what confidence threshold do you have in mind for escalation? Thanks, Differenz


a support flow in n8n where claude and gemini check each other's replies sounds like a good one, and i can build it for you i can set this up end to end and get it working on your real chat and social messages i'll stay on it until it runs the way you want it to which helpdesk or ticket tool are you using right now? happy to go through the rest with you in chat Sela
内容来自 Freelancer News,版权归原作者所有。
阅读原文 ↗