AI Agents

Standard, reasoning, autonomous, multimodal, and specialist AI agent nodes.

Overview

The AI Agents group holds nodes that call a large language model to reason over upstream data, generate content, or take autonomous action within a pipeline. Each node type trades off autonomy, cost, and predictability differently — from a single-turn Standard call to a fully Autonomous agent that plans its own steps.

Writer Engine routing applies here too

AI Agent nodes route through the same Writer Engine setting (Local LLM / Web LLM / Ollama / API) as the rest of the app. Whichever engine is selected in Settings → Agentic is used when the node executes.

Screen preview

A quick look at the screens. Use the toggle in the corner of each preview to switch between the dark and light themes.

app.totalapp.app/ai-agents

Agent Types

Standard

Reasoning

Autonomous

Multimodal

AI agent types

app.totalapp.app/workflow-editor

AI Content Pipeline

Prompt: "Generate a blog post about sustainable packaging trends in 2026"

Text Input

util · source

Done

↓

Blog Writer

dw-content · AI

Running

↓

Image Generator

ig-generate · AI

Pending

↓

Email Publisher

output · send-mail

Pending

↓

Slack Notifier

output · send-message

Pending

↓

Analytics Logger

output · log-event

Pending

AI content pipeline on the workflow canvas

Nodes in This Group

Node	Description
Standard	A single-turn LLM call: send a prompt (with upstream data injected), receive a response.
Reasoning	Uses an extended-reasoning model configuration for tasks that benefit from step-by-step thinking before answering.
Autonomous	An agent that can plan and execute a sequence of steps on its own, rather than producing a single response.
Multimodal	Accepts image or file input alongside text in the prompt.
Specialist	A pre-configured agent tuned for a specific task type (e.g. code review, data extraction).
Domain-Specific	A pre-configured agent tuned for a specific business domain (e.g. legal, finance, HR).
Custom	A user-defined agent built from a saved persona and tool configuration in Agent Persona Editor.
Document Analyzer	Extracts structured information from an upstream document.
Map Columns	Uses an LLM to map upstream column names/values onto a target schema.
Sentiment Scorer	Scores the sentiment of upstream text (positive/neutral/negative or a numeric score).

Usage Tips

Start with Standard

Reach for the Standard node first for most single-step generation or extraction tasks. Move to Reasoning or Autonomous only when the task genuinely needs multi-step planning.

Custom agents come from Agent Persona Editor

Build and save a persona (system prompt, tool access, model choice) in AI Orchestration → Agent Persona Editor, then reference it from a Custom node in the workflow.

Frequently Asked Questions

Can an Autonomous agent call other nodes in the workflow directly?

No. An Autonomous node plans and executes its own internal steps using the tools it has been given access to, but it does not reach outside the node to invoke other workflow nodes — data still flows through normal node connections.

What happens if the configured Writer Engine is unavailable when the node runs?

The node fails with an error rather than silently falling back to a different engine — check Settings → Agentic to confirm the selected engine (e.g. Ollama) is running before executing the workflow.