Zenphi adds Claude as governed workflow agent in Google Workspace

Zenphi launched the Claude Text Agent on July 3, 2026, giving Google Workspace teams a governed way to run Anthropic’s models inside business workflows instead of only in chat. The move expands model choice in Workspace and lets organizations route reasoning-heavy tasks to Claude while keeping execution logged, auditable, and connected to downstream systems.

Why it matters: - Zenphi is moving Claude from a chat assistant into automated business processes inside Google Workspace. - The setup gives teams model choice in an environment where Gemini is the default AI. - Enterprise users can send high-reasoning tasks to Claude without leaving their existing workflow stack. - The approach is designed for governed execution, with each run scoped, logged, and audited.

What happened: - Zenphi announced the Claude Text Agent on July 3, 2026. - The agent runs Claude models as configured agents inside production workflows. - The launch is part of Zenphi’s AI Agent suite, which already includes agents for other model providers. - Google Workspace teams can use the agent for tasks such as reading contracts, classifying documents, and returning structured data.

The details: - Zenphi is built natively for Google Workspace. - The Claude Text Agent is designed as a first-class agent, not just a connector that sends prompts to an API. - Workflow designers can set system-level instructions to define the agent’s persona and behavior across runs. - Documents and images can be passed directly to the agent without a separate pre-processing pipeline. - Users can switch outputs between raw text and valid JSON. - Downstream workflow logic can act on the agent’s output automatically. - Every execution reports total, input, and output token counts. - Token reporting gives finance and IT teams per-run and per-workflow cost visibility. - Zenphi workflows use deterministic logic for conditions, routing, and data operations. - The model is invoked only at steps that require reasoning. - In a legal operations example, the Claude Text Agent extracts liability caps, renewal terms, and termination conditions from incoming contracts. - The workflow validates the JSON output, writes it to the system of record, and sends exceptions to a human reviewer. - The process removes manual copy-paste between a chat window and business systems.

Between the lines: - The launch reflects a broader shift from AI as a standalone assistant to AI embedded in operational software. - Zenphi is positioning workflow control, auditability, and cost tracking as the enterprise differentiators, not just model quality. - The emphasis on model choice signals that Google Workspace customers may want to route work by task instead of relying on one default model. - The setup also suggests enterprises want frontier models where judgment matters, but deterministic software where repeatability matters.

What’s next: - Zenphi will likely keep expanding its AI Agent suite with additional model options. - Workspace teams can expect more workflows that combine governed automation with model-based reasoning. - Organizations that already run processes in Google Workspace can adopt Claude without redesigning their operating environment.

The bottom line: - Zenphi is turning Claude into a governed workflow worker, not just a chatbot, and that could make model choice more practical for enterprise teams already living in Google Workspace. - More information is available at zenphi.com.