Runtype
Agents

What are Agents?

Agents are agentic AI workflows that autonomously decide which tools to use and how to accomplish goals, unlike Flows which follow predetermined steps.

How Agents differ from Flows

Flows execute a fixed sequence of steps you define. Agents reason about problems and choose their own actions.

Flows

Agents

You define each step

Agent chooses actions

Predictable sequence

Dynamic reasoning

Best for defined workflows

Best for open-ended tasks

Faster, more deterministic

Flexible, more autonomous

How Agents work

You give an Agent:

  • Goal: What to accomplish

  • Tools: Functions it can call

  • Context: Information to work with

The Agent then:

  1. Analyzes the goal

  2. Decides which tool to use

  3. Calls the tool with appropriate parameters

  4. Reviews the result

  5. Repeats until the goal is achieved

Agent in action

Sarah, a product manager at a interior design company, needs to compile competitive intelligence reports every Monday morning. Her stakeholders want summaries of what competitors shipped last week—but the information is scattered across blog posts, release notes, and documentation sites.

She tried building a Flow, but hit a problem: each competitor publishes updates differently. Some announce on their blog. Others only update documentation. A few use Twitter threads. A fixed sequence of steps couldn't adapt to these differences.

Instead, Sarah builds an Agent.

She provides:

  • Goal: "Research what [competitor name] shipped last week and summarize key features"

  • Tools: Exa web search, Firecrawl scraper, a custom JavaScript function that formats markdown reports

  • Context: The competitor's domain and a list of keywords to watch for

The Agent runs:

When Sarah triggers the Agent with "Example Competitor" as the competitor, it autonomously decides:

  1. Search first. Calls Exa to find recent articles mentioning "Example Competitor" and "launch"

  2. Found a blog post. Calls Firecrawl to extract the full content from examplecompetitor.com/blog

  3. Spot more references. The blog post mentions a new docs section. Calls Firecrawl again to scrape that page

  4. Synthesize findings. Analyzes both sources, identifies three new features

  5. Format output. Calls the custom formatting tool to generate a markdown report

  6. Done. Returns the final summary to Sarah

Sarah never specified these steps. The Agent chose its path based on what it found. When it ran the same Agent for a different competitor the next week, it took a different route—going straight to documentation because search returned nothing recent.

That's the difference: Flows execute your plan. Agents make their own.

When to use Agents

Use Agents for:

  • Complex, multi-step research tasks

  • Open-ended problem solving

  • Tasks requiring adaptive decision-making

  • Workflows where the path varies by input

Use Flows for:

  • Predictable, repeatable workflows

  • Tasks with fixed steps

  • When you need guaranteed execution order

  • Performance-critical operations

Start with Flows. Add Agents only when you need autonomous reasoning. Agents are more expensive and harder to debug than Flows.

Available tools

Agents can use various tool types:

  • Built-in tools: DALL-E, Exa search, Firecrawl scraping

  • Custom tools: JavaScript functions you write

  • External tools: HTTP APIs

  • Flow tools: Run existing Flows as tools

  • MCP tools: Connect to Slack, Google Workspace, Linear, GitHub, or custom MCP servers

See Agent tools and Tools overview for details.

Next steps

Was this helpful?