From Workflows to Agent Orchestration
Mike Knoop, co-founder of Zapier, on Zapier's LLM-powered future
This points to automation moving from rule writing to worker management. Zapier started as a place to connect apps with rigid if this then that logic, but the company is explicitly steering toward software that can take a job description, use connected apps to carry it out, and improve through correction over time. That shifts the product from a workflow builder into a layer for supervising software labor across thousands of SaaS tools.
-
The first concrete step is setup. Zapier described its Natural Language Actions API as turning plain English instructions into app actions, with preview and user overrides so people can approve risky choices like which Slack channel to post in. That is the bridge from configuring fields manually to teaching software what good output looks like.
-
The bigger change comes in execution. Zapier said the next phase is software making smart choices while running, not just during setup. That is much closer to how a human assistant works. The manager gives goals, connected accounts, and guardrails, then checks logs and corrects mistakes after the fact instead of specifying every branch in advance.
-
Competitors show the same market split. Workato is packaging governed enterprise agents with skills, knowledge, and observability, while n8n emphasizes technical teams building AI workflows and agents with human review and natural language workflow generation. The shared pattern is that automation tools are becoming systems for deploying and supervising agents, not just drawing boxes between apps.
The category is heading toward agent orchestration as the new automation interface. The winners will be the platforms that already hold user trust, app connections, execution logs, and control points, because those assets let them train, monitor, and bill for software workers at scale instead of just selling seats for a visual builder.