New Claude Agent moves frontier reasoning out of the chat window and into business processes — giving Google Workspace teams model choice on Gemini’s home turf
DOVER, DE, UNITED STATES, July 3, 2026 /EINPresswire.com/ — Most enterprises have met Anthropic’s Claude in a chat window: an assistant that drafts, summarises, and answers when asked. Today, Zenphi — the AI automation platform built natively for Google Workspace — announced the Claude Text Agent, which gives Claude a different kind of job. Claude models now run as configured, governed agents inside production business workflows: reading contracts, classifying documents, and returning structured data that downstream process logic acts on automatically — with every run scoped, logged, and audited.
The launch also marks a notable moment for the Google Workspace ecosystem, where Gemini is the default AI. The three billion users working in Google Workspace increasingly expect model choice — the ability to route each task to whichever model handles it best. With the Claude Text Agent joining Zenphi’s AI Agent suite alongside agents for other leading model providers, Google Workspace teams can now assign their highest-reasoning workloads to Claude without leaving the environment their processes already run in.
From Connector to Agent
In most general-purpose automation tools, Claude is available as a connector: a step that sends a prompt to the API and maps the response onward. That works for simple text tasks. It leaves the harder questions — how the model should behave, how its output stays machine-readable, what each run costs — for the workflow builder to solve by hand. This mental model is unacceptable for enterprise workflow automation.
Zenphi built the Claude Text Agent as a first-class agent instead. Designers define system-level instructions that fix the agent’s persona and behaviour across every run. They pass documents and images to it directly, with no pre-processing pipeline. They toggle output between raw text and valid JSON, so results flow straight into downstream logic, lookups, and integrations. And every execution reports total, input, and output token counts — per run, per workflow — so finance and IT know precisely what each AI-powered step costs.
Within the Zenphi enterprise automation platform, the architecture around the agent matters as much as the agent itself. Zenphi workflows use deterministic logic for conditions, routing, and data operations, so the model is invoked only at the steps that genuinely require reasoning. Claude does the reading; the workflow guarantees the process.
“Most AI tools start with the model and then go looking for a business problem to point it at. Zenphi’s AI automations are different,” said Vahid Taslimi, CEO of Zenphi. “The processes a company runs on have to be repeatable, predictable, and cost-efficient — those requirements come first, and the workflow simply embeds AI wherever it delivers maximum value. The Claude Text Agent is exactly that: a frontier model doing a defined job, inside a process that was already built to run reliably.”
What It Looks Like in Practice
A legal operations team routes incoming contracts through a Zenphi workflow. The Claude Text Agent reads each file, extracts the clauses the team tracks — liability caps, renewal terms, termination conditions — and returns them as structured JSON. The workflow validates the output, writes it to the team’s system of record, and routes exceptions to a human reviewer. No employee copies contract text into a chat window; no reviewer re-keys the model’s answer into another system.
Key capabilities of the Claude Text Agent:
• Model selection — choose from supported Claude models, to balance speed, cost, and intelligence per workflow step.
• System instructions — granular, system-level behaviour and persona control, consistent across every run.
• File input support — pass documents or images directly for analysis, classification, or summarisation.
• Structured and JSON outputs — toggle between raw text and valid JSON, ready for downstream logic.
• Token usage analytics — total, input, and output token counts on every run for precise cost monitoring.
About Zenphi
Zenphi is the only no-code AI workflow automation and AI agent platform built natively for Google Workspace, processing over 1.5 million AI-powered workflow tasks per month for organisations including Google, Gordon Food Service, Action Behavior Centers, and Emerson College. The platform enables IT, HR, finance, legal, and operations teams to automate complex workflows and deploy governed AI agents without engineering resources. Zenphi is HIPAA compliant, ISO 27001 certified, GDPR compliant, CASA Tier 2 verified, and a certified Google Cloud Partner. More information is available at zenphi.com.
Ana Bibikova
Zenphi
email us here
Visit us on social media:
LinkedIn
YouTube
Legal Disclaimer:
EIN Presswire provides this news content “as is” without warranty of any kind. We do not accept any responsibility or liability
for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this
article. If you have any complaints or copyright issues related to this article, kindly contact the author above.
Media gallery
