Unleashing the POWER of Gemini 2.0 Flash in n8n

Join me on Saturday, December 28th, at 9:00 AM CDT for a livestream as a part of me kicking off my new mini series on my entire process for building AI agents!

This is the prototype phase of my AI agent build process. Check out this video (publishing the 22nd at 7:00 PM CST) where I reveal my AI agent blueprint that this is a part of:

[Goldbay]

I mostly live on the road in airbnbs in california, nevada and arizona. I am watching this from Guadalajara mexico and AI and IDEs have totally changed my life and direction. I am devouring content to become an expert, Thanks for your vids!

[llucis-v]

this is the best n8n tutorial/intro i found for my level (never used n8n before) and the comments/observations/answers are top notch! thank you for this video!

[nostalgicnow]

Thanks for making this as a cybersecurity consultant and software dev. This is useful and fun and will be rewarding in the near future

[kuyajon]

Keep up the good work Cole, love from Manila Philippines

[hemantrathod]

hey, there is no "field to return" section in "call n8n workflow tool....how do i overcome this problem?

[Thecoolestcouple]

@Cole could you please enable the Live Chat Replay. Can't see any of the comments or links that you posted. Thx.

[SuaveBet]

The "field to return" field is missing in my "Call n8n Workflow Tool." Is there something I need to do to enable that field?

[SouthbayJay_com]

Hey Cole! Fantastic video!! I missed the live, it was too early for my time zone in San Diego, but I rewatched it twice!! So much great info!! Keep up the awesome work! Jay

[josejunior]

Hello, could you enable the YouTube functionality that adds the possibility to select other dubbed audio tracks for this video? 🙏🙏🙏

[mfauzi]

Hi. Do we need to list the tools in system prompt? Or as you said it already baked when we give it in tool name and description? Because my agent doesn't recognize the tool at all.

Edit: It working now. Weird. But it not consitently working. I use Gemini 1.5 flash

[olb]

Is it possible to link n8n with langfuse?

[skyblue]

Hey my comment disappeared. I tried to install the self hosted ai from your video.

I'm getting an n8n encryption key error which i cannot resolve.

Can anyone point me to a guide?

[rocketPower]

one thing to note is that in the code blocks, I was going mad cause I kept getting errors in the workflow tool. If you played around with blocks too much, createing a new one will suffix a digit to the block. So my 2nd code block was JS ended with:

return { structure: formatRepoStructure($('HTTP Request').all()[0].json.tree) };

But the block was called "HTTP Request1", and that caused it to fail! I thought the names of the blocks were just for human readability, but they serve a code purpose too

[rocketPower]

I managed to get away with llama 3.2:3b for this agent.

[alx]

Also n8n literally sucks when it comes to tracing of what kind of things being sent to LLM / responded by LLM - the worst tool ever. But your video is great, thanks for it

[Minotaurus]

I like his videos about n8n, but this here is chaos to me. Building agents was so easy, but here he starts to make it difficult. His road map threatened this already. No code now seems to be way more difficult than coding. :-)

[alx]

n8n "AI Agent" node is the sloppiest implementation of agent I ever seen. I've been tracing what kind of LLM calls does it do, what kind of things LLM being asked of - and it's terrible. It looks like it was made by some under-graduate student. This only works if you plug it with some really clever model, like qwen-2.5.-32b or any proprietary one. Implement your own agent - it will be much more robust, quick and reliable

[citizen_cicero]

Brother, love ya but drop the chin strap beard. Either grow in the beard or stick with a goatee and sideburns but don't connect them.