Hello dear friends, Thank you for reading this, let me get into it: I have a workflow that takes a form (onboarding form) and creates a market research and creative brief out of it. I do this with HTTPS and DeepSeek. Using deepseek means I have to use JSON to add the prompt (as I know of) and I don't know nothing about JSON. usually the prompting goes well, If I go from the form to deepseek right away. But if I have another node in the middle (for example: form - drive - https) and I add info to the prompt from the form it says an error (JSON parameter needs to be valid JSON) And right now, I have the same problem, as I added to the end of the workflow an other DeepSeek node (https) where it takes the market research document and writes an Ad Hook from it. The point is the "Get Document" is before the deepseek node, and it runs smoothly until I add any expression or data from previous nodes. "role": "user", "content": "Market Research: {{ $json.content }}" What is the JSON problem and how can I reliable always reference data without JSON error? I would actually pay for someone's help

In this video, I’ll show you how to connect DeepSeek’s newest reasoning model, DeepSeek R1, directly in n8n—no OpenRouter required! Because this model isn’t built for tool/function calling, we want to use DeepSeek R1 as a planning agent. It’s job is to take the human request, and think through the steps to achieve the result. It will then feed this plan of action into a tools agent that can use any other model like GPT 4o or Claude 3.5 Sonnet.

Hey everyone - new to the group! Really great to see this community thriving. Just built my first automation yesterday and safe to say I'm floored with the potential of creating your own automations! I've attached a screenshot my workflow using telegram, my email and open AI. My question is, is N8N safe? If I give access to company data (ie. emails), is there inbuilt security protocols that I can rely on? I want to bring this to my employer but I want to understand what the implications are first

Hello everyone! :) I'm excited to be part of this community! Is there anything I can help with? I'm currently working on building a matching workflow. I have two tables in Supabase: - Job Descriptions (documents/content) - Candidates (kandidaten/content) I aim to match these two datasets effectively, but the data volume is quite large. Any advice, tips, or best practices for handling large-scale matching workflows would be greatly appreciated. Looking forward to learning from you all!

I don't know if this idea will take but it might be valuable to start an architecture discussion thread. At some point, an AI agent could scrape it and document best practices 😎 In either case, I have a basic question. The screenshot shows a simple workflow that pulls contacts from a Google sheet and emails them based on a user chat prompt. If the user asks to email a contact whose information is not listed, the agent is already smart enough to ask for the email address but not yet smart enough to save the new information in the contact list. My question is, where is the best place to make this sort of decision? I could add to the System Message within the AI Agent, but I think that would be prone to error, and possibly bad practice. Is an "if" statement appropriate here? What am I missing? Any advice is greatly appreciated