Skip to main content

Updated Sep 30, 2026

Open in ClaudeOpen in ChatGPT

AI Agents

AI Agents are configurable automations you build in FunnelStory: pull data from your workspace or connections, call an LLM where you need judgment or language, then take action — Slack, email, tasks, CRM updates, datasets, and more. Reach for one whenever a built-in feature's fixed rules aren't enough and you need branching logic, a query across your workspace, or several actions chained from one run.

The shape: trigger, step graph, action​

Every agent is a directed graph of steps, wired together with a trigger that decides when a new run starts — on a schedule, when an event or signal occurs, when a query returns matching rows, or only when someone starts it manually or from chat. Steps call functions (CALL) or run an LLM for multi-step reasoning (AGENT), passing results forward through variables. See AI Agents overview for the full mental model, including the operation types and how runs and events work.

Two ways to build​

You author an agent either on a Canvas — set the trigger, add blocks, configure each step in a side panel — or through Vibe Coding, describing what you want to an AI assistant connected over MCP, which reads FunnelStory's own flow authoring guide and builds or edits the agent for you. See Getting started and Vibe coding.

AI Agents vs. built-in and Outcome Agents​

Not every agent in FunnelStory is one you build. Built-in and Outcome Agents are built by FunnelStory around a specific job or goal, with nothing to author — the Needle Mover investigation agent and Outcome Agents are examples. AI Agents are the general-purpose counterpart: the graphs are yours to design, for whatever your team needs. See Built-in and Outcome Agents for the full comparison.

Last updated on