IT Brief Asia - Technology news for CIOs & IT decision-makers
Flux result 5716a27b 3cfe 4e12 bd3f 9c858330c925

Zapier expands AI governance controls for enterprise users

Fri, 24th Apr 2026 (Yesterday)

Zapier has expanded its AI governance offering for enterprise users, extending policy controls across workflows, AI agents, connected assistants and software built with its SDK.

The update gives IT and security teams a single policy layer across the places where AI tools are created and used inside organisations. That includes no-code workflows, AI agents, assistants connected through the Model Context Protocol, and applications built with Zapier's software development kit.

The move comes as companies try to expand their use of AI while tightening control over data access, application connections and system changes. Zapier also published survey findings from 200 enterprise executives. Among them, 93% said AI initiatives in their organisations at least occasionally fail to reach production because of governance constraints, while 94% said governance needs to move beyond static policies and approvals to a system embedded in how AI is built and run.

"Building has gotten very easy, very fast. And for most organizations, new ideas are already outrunning the guardrails around them," said Wade Foster, co-founder and chief executive officer of Zapier.

"If building is everywhere, governance has to be everywhere too," Foster said.

New controls

The additions include app access controls that let administrators decide which applications teams can use, with settings applied by workspace, team or individual user. Those controls are enforced across Zapier's editor, agents and MCP connections.

Zapier has also introduced action restrictions, allowing businesses to define which actions users can take inside an application. For example, a sales employee may be allowed to read and update contacts in HubSpot but not delete them.

Another feature, managed app connections with domain restrictions, is designed to give companies more control over how employees connect third-party services to Zapier. The aim is to reduce the risk of data moving through personal accounts and to prevent workflows from failing when a member of staff leaves.

Model routing

Zapier is also adding what it calls Bring Your Own Model. The feature is intended to let customers decide where company data and knowledge are processed by routing Zapier agents and knowledge processing through their own infrastructure, starting with AWS Bedrock.

For larger organisations with multiple departments, workspaces will let each team operate in a separate environment with its own applications, connections and policy settings. The feature is expected to reach general availability by the end of the second quarter.

The release also includes log streaming and asset history. Zapier said these tools show what the system touched, what failed and when, with information streamed in real time to services such as Datadog, Splunk or another security information and event management platform used by a customer's security team.

Agents and SDK

At the same time, Zapier's Agents product is now generally available with enterprise MCP support. That means businesses can use agents and connect them to internal systems both inside workflows and outside the Zapier platform while keeping the same governance controls applied elsewhere in the product.

The Zapier SDK has also entered open beta. It is intended to let developers connect agents to applications, workflows and Zapier's wider catalogue of integrations from different environments, while remaining subject to enterprise policy controls.

The broader announcement reflects a shift in how software companies are positioning AI management tools for large organisations. Instead of treating governance as a separate approval step, vendors are increasingly building policy enforcement into the day-to-day systems used by employees, developers and business teams.

Zapier framed its latest release around that shift, arguing that the old model of governance was built for a smaller group of technical builders working in a limited number of controlled systems. Its updated approach is aimed at a landscape in which non-technical staff, departmental teams and external tools can all create or trigger AI-driven processes.

All of the governance features announced are now available to Zapier customers, with workspaces due to follow in general availability later in the quarter.