Yonathan Cohen on Streamlining Automations with LLMs, Bypassing Traditional Workflow Tools

In a recent LinkedIn post, Yonathan Cohen explores a new paradigm for business automations, suggesting that Large Language Models (LLMs) like Claude can now handle many tasks previously requiring complex workflow tools such as n8n, Zapier, or Make. Cohen argues that the perception of LLMs as mere chatbots is outdated, positioning them instead as an emerging “action layer” over existing business tools.

LLMs as an Action Layer

Cohen highlights the potential of LLMs to directly integrate with and control various applications, a capability he refers to as leveraging “MCPs” (presumably Machine Control Protocols or similar). This integration allows users to initiate multi-step workflows through simple natural language prompts, rather than building intricate sequences in dedicated automation software.

He elaborates on this shift:

“Most people still think Claude is a chatbot. But it’s becoming an action layer over your tools.”

This perspective challenges the conventional approach to automation, where setting up even a seemingly simple task can involve numerous steps within a workflow builder.

Simplifying Complex Workflows

Cohen contrasts the traditional method with his proposed LLM-driven approach. He points out the inefficiency of building lengthy workflows for routine tasks.

“Most people build a 20-node workflow in n8n to send one personalized email,” Cohen states, illustrating the over-engineering he believes is common.

Instead, he proposes a more streamlined process:

“I type: ‘Research this company, check my calendar, and send them an email with my availability.’ One prompt but same outcome.”

This single-prompt method, according to Cohen, achieves the same result as a complex, multi-node workflow, significantly reducing the time and effort required.

The Role of Connected Tools

The effectiveness of this LLM-based automation hinges on the model’s ability to connect with essential business tools. Cohen lists several key integrations:

• Your email
• Your calendar
• Your CRM
• Your browser
• Your files

Once these connections are established, the interaction with the LLM transforms from a conversational exchange to a command-and-control interface for executing business processes. As Cohen puts it, “Once connected, you don’t ‘chat’ with Claude. You run workflows by talking to it.”

When Traditional Tools Still Shine

While advocating for the power of LLMs in simplifying many automations, Cohen acknowledges the continued relevance of traditional tools like n8n, Zapier, and Make. He clarifies that these platforms remain superior for specific use cases.

“n8n, Zapier, and Make still win for automations that run 24/7 without you.”

This distinction is crucial: LLM-driven automations, as described by Cohen, are best suited for tasks that are initiated by a user prompt and do not require continuous, unattended background operation. For background processes, scheduled tasks, or complex, long-running jobs that need to operate independently, the established workflow automation platforms still hold the advantage.

In conclusion, Yonathan Cohen’s insights suggest a significant evolution in how business professionals can approach automation, leveraging the advanced capabilities of LLMs to simplify workflows and enhance productivity for prompt-initiated tasks, while recognizing the enduring value of traditional tools for continuous background operations.