The useful question is what changes for users, developers or buyers, and whether the announcement stays industry context or becomes something people can actually use.

Zapier’s new AI tools allow users to add AI steps to automated workflows, called Zaps, without running the entire process through an AI model. These steps, powered by models from OpenAI, Anthropic, or Google, can summarize data or provide context where needed, reducing costs by charging only for AI usage. Users can ground AI steps in company knowledge from connected apps like Google Drive or Notion, or enable agentic behavior for multi-step tasks. Nearly 700 AI apps also integrate directly with Zapier for additional flexibility.

Zapier Tables introduces AI fields and AI enrichment to streamline data management within workflows. AI fields apply a prompt to every row in a column, regenerating outputs when referenced data changes, while AI enrichment populates multiple columns based on a single prompt. Both features are included in all plans, including the free tier, and eliminate the need for external spreadsheets or complex formulas.

For teams using AI tools like ChatGPT or Claude, Zapier offers secure integration paths via the Zapier SDK or CLI. The SDK, currently in open beta, enables building within coding agents with loops and conditionals, while the CLI allows exploring app capabilities and running actions from a terminal. Direct API calls remain ungoverned, but pre-built actions are secure. Installation guides are available for further details.

Zapier also provides AI-powered tools to assist in building and troubleshooting automations. Zapier Canvas maps workflows with AI diagrams to identify bottlenecks, while Zapier Copilot generates cross-product systems from plain-language prompts. Code steps can auto-generate JavaScript or Python, and AI Guardrails screens content for PII, prompt injection, or toxic language before it reaches business systems.