The ideal team of the future isn’t 10 people with specialized skills. It’s one person with a team of agents. That’s the vision Zapier CEO Wade Foster laid out in a recent conversation with Adam Brotman and Andy Sack on the AI First podcast.
As AI reshapes how work gets done, Foster argues that workflows and agents aren’t competing approaches—they’re complementary tools that work best together. The companies that figure out how to combine them will have a massive advantage.
Table of contents
Workflows vs. agents: Pick the right tool for the job
There’s a lot of confusion about what counts as “automation” versus “agents.” Foster breaks it down simply. A workflow is deterministic, which means you define steps one through five, and the system executes them in order. With an agent, you give it a goal and tools, and the LLM decides how to accomplish the task.
Neither is inherently better. They just have different trade-offs.
Agents have really two tricky challenges compared to workflows. They are generally less reliable—they’re probabilistic. And the second thing an agent struggles with is it’s more costly. They’re burning lots of tokens to go figure out what that task is.
Wade Foster, CEO at Zapier
Foster’s advice? Run more things as workflows if you can. That doesn’t mean your workflow can’t have AI components. But if you care about cost and reliability, you want something closer to deterministic code than pure agent inference.
Where agents shine is in the creation process. When you’re prototyping, iterating, and not sure exactly what you want, agents are invaluable. But once you’ve figured it out, you lock it into a workflow.
Why agents should back up your workflows
Workflows are reliable, but they’re also brittle. An API changes, a site goes down, and suddenly your automation breaks with no way to recover.
What if the workflow is backed by an agent? When it fails, the agent is then called in to troubleshoot and say, ‘This error message came back. I wonder why. Here are three ideas I have for how I can maybe go fix it.’
Wade Foster, CEO at Zapier
The answer isn’t workflow or agent—it’s both. Zapier is building systems where agents create workflows, and agents back up workflows when they fail. The human’s job shifts from fixing broken outputs to fixing the inputs that caused them.
Foster compares it to modern manufacturing. When a machine produces a defective widget, you don’t manually fix the widget, you fix the machine. Humans add value in an AI-powered workflow by designing systems, tuning instructions, and improving the process when outputs go wrong.
The future org chart: Humans plus agent teams
Most organizations today are structured by function, like Marketing, Sales, or Engineering. These teams exist to provide specialized skills for projects, but LLMs increasingly represent those skills themselves.
What is the ideal team? It’s a domain expert who knows the problem they’re trying to solve, knows the customer really well, and can now assemble that team of agents to go tackle those problems.
Wade Foster, CEO at Zapier
Foster sees organizations evolving toward loose collections of humans, each managing their own agent teams to accomplish goals. Instead of 100 people split into 10 functional teams, you might have 50 domain experts—generalists who understand the business—each orchestrating agents that handle the specialized tasks.
This has implications for hiring. Zapier now evaluates every candidate on AI fluency, with a rubric customized by function. We’re looking for people who are curious, actively using the tools, and “trying to live in the future.”
For people entering the workforce, Foster sees this as an opportunity. Platform shifts level the playing field—someone with 20 years of experience may have skills that are less relevant for the future. If you master tools like Zapier, Claude, or ChatGPT, you can rocket your career forward.
The key traits: Be entrepreneurial, understand customer problems, and start building tools to solve them.
This article is based on Wade Foster’s appearance on the AI First podcast with Adam Brotman and Andy Sack. Watch the full conversation for more on Zapier’s MCP server, how Foster uses a “cabinet” of AI advisors for decisions, and why Opus 4.6 is his daily driver.
Zapier is the most connected AI orchestration platform—integrating with thousands of apps from partners like Google, Salesforce, and Microsoft. Use forms, data tables, and logic to build secure, automated, AI-powered systems for your business-critical workflows across your organization’s technology stack. Learn more.
