Artificial Intelligence
Artificial Intelligence Wrk Actions invoke leading AI models for classification, extraction, summarization, and generation. AI steps fit into Wrkflows like any other Action: with inputs, outputs, and downstream logic.
What it is
Each AI Wrk Action calls a specific model or capability from a supported provider (OpenAI, Anthropic, Gemini, or Perplexity). You pass prompts, files, or structured inputs and receive text, JSON, or other outputs you route to the next step in your Wrkflow.
Good for
- Text classification and routing by category or sentiment
- Content generation: draft responses, summaries, or rephrased copy from structured inputs
- Summarizing long documents or conversation threads
- Vision and Q&A: ask questions about images using OpenAI Vision
- Agentic workflows: the Wrk AI Agent can select tools and invoke Wrkflows to complete a task autonomously
What it does
- Generates chat completions across OpenAI, Anthropic, and Gemini model families
- Rephrases text, extracts keywords, and moderates content
- Converts speech to text and text to speech
- Performs web-grounded search through Perplexity
- Runs an autonomous Wrk AI Agent that can call Wrkflows and tools to complete multi-step tasks
Go deeper
Reference
Related concepts
- AI in Wrkflows: common patterns and when to add Human-in-the-Loop review
- Human-in-the-Loop