Understanding the future: AI Agents or automation workflows (or both!?)
This should help you understand the differenced between AI Agents & automation workflows (they are complimentary)
Making sense of the different ‘AI agents’ currently avaible - Convergence AI, AgentGPT, LangChain, the list goes on….
At first glance, they seem like the future: smart, autonomous bots that can browse the web, plan tasks and ‘think’ their way through problems.
Here’s the catch 22: they’re clever, but not very reliable.
Agents today are slow, they get stuck and are expensive to run. Great for research or prototyping. Not great if you need consistent results every day.
These are not to be confused with tools like n8n - This is ‘workflow automation’ that uses triggers from webhooks, APIs, and now even LLM’s built in.
These are less autonomous and need manual programming/setup, but more stable, repeatable, and still super useful.
The sweet spot?
-
Use AI agents to think (e.g., generate ideas, make decisions)
-
Use workflow automation, with tools like n8n, to do (e.g., send emails, update sheets, post content)
You can plug LLM’s into n8n with HTTP requests or the new “AI Agent” node.
So yes - agents might be the future, but automation featuring AI Agent intergration (a combo of both) is the present. I am certianly utilising both.
Hey, I just published a really sweet guide on how to self host n8n with a Hetzner VPS (The best option for price, performance and security, click here)
That’s all, folks!
Dom’s Content