What is Chase Agents?

An automation platform for AI-powered workflows

Chase Agents is an automation platform that lets teams build, schedule, and run multi-step AI workflows without writing backend infrastructure. Connect your APIs and tools, describe the work once, and Chase Agents executes it autonomously — on schedules, webhooks, or on-demand.

Two main building blocks

AI Agents

An AI agent is a conversational assistant powered by a custom system prompt. You give it a title and a system prompt, and users chat with it in the Chase Agents UI. The agent has access to all your workspace connections and can query live data, answer questions, and take actions during the conversation. Agents are interactive — they respond to what a user says.

Automations

An automation is a saved multi-step workflow that runs without user interaction. It can be triggered on a cron schedule, at a fixed interval, by a one-time timer, via a webhook from an external service, or by a direct API call. Each automation is a sequence of typed steps that execute in order, passing data between them. Automations are autonomous — they run whether or not anyone is watching.

Step types

Automations are built from typed steps. Every step has a type that determines what it does:

  • call_api_endpoint — make HTTP requests to any connected REST API
  • run_sandboxed_code — execute Python for data transformation, calculations, and AI-powered text generation
  • use_mcp_tool — call a tool on a connected MCP server
  • if, switch, loop, merge — conditional branching and iteration
  • generate_pdf, generate_word — produce formatted documents
  • combine — merge multiple files of the same type into one
  • wait — pause execution for rate limiting or timing
  • trigger_automation — run another saved automation
  • seek_human_approval_for_next_step — pause and email a human to approve before the next action proceeds

Connections: the foundation

Both agents and automations operate through connections. A connection is a pre-configured API or MCP server that your workspace has authenticated. Credentials are stored encrypted and referenced in steps via template variable names (written as the variable name wrapped in dollar-sign curly braces). The system replaces those references with real values at runtime — your automation logic never handles raw credentials.

When to use agents vs. automations

Use an AI agent when you need an interactive, conversational experience where a user asks questions and the agent responds with real data. Use an automation when you need a workflow that runs repeatedly and predictably without human involvement — on a schedule, in response to an event, or triggered by another system.