Zurich, Switzerland — On the 21st of October 2026, ues.io will release the 2026.10 Gotthard Release, transforming the platform into an enterprise MCP application platform for building and running governed AI agents. Banks, insurers, wealth managers and public-sector teams can now build applications that work natively with AI agents, keeping every agent within the same permissions, approvals and audit trails that apply to their people. Applications, data and AI agents now run under one permission model, with swappable models, self-hosting, and Swiss or sovereign EU hosting.
Control, not just connection
Most enterprises can already connect an agent to their systems. The harder question is control: which records the agent may read, who approves its writes, and how to prove afterwards what it did. The Gotthard Release answers that on the platform where the application itself is built.
Named after the Swiss Alpine pass that connects people and economies, the Gotthard Release stands for overcoming boundaries. It connects AI agents with enterprise systems and ensures they operate within a robust governance framework with clearly defined permissions, approval processes and comprehensive audit trails.
What is new in the Gotthard Release
Governed agent orchestration with swappable models — Agents built on ues.io run under native MCP orchestration. They can move between commercial LLMs and self-hosted SLMs without changes to the application.
MCP-Apps — ues.io acts as both an MCP server and an MCP client. Any ues.io application can expose its data collections, views and workflows as MCP tools to external AI agents, including Claude, ChatGPT and in-house enterprise agents. What data and functions those tools can use depends on the permissions of the individual user.
Coding agents that build within guardrails — ues.io coding agents create and change applications through the platform’s own metadata: data models, views, workflows and permission sets. Every code change can be versioned in Git, and sites can be rolled back to an earlier release.
Multi-tenancy and permission sets — Permission sets decide what people and AI can access. Applications and agents can be built once and deployed to many tenants with isolated data.
“Every enterprise can now connect an agent to its systems. What regulated firms ask us is whether they can control it. In the Gotthard Release, an agent is a user like any other: it has an identity, it sees only what that identity may see, and everything it does is on the record.”
Gregg Baxter, CEO, ues.io
Governance without a second security layer
On ues.io, users, agents and MCP tools are governed by the same permission sets and record-level security. There is no separate agent security layer to keep in sync.
User identity passthrough — An AI agent acts on behalf of a clearly identified person and can only read the records that person has permission to read.
Audit trails — Agent executions, tool calls and data changes are logged and linked to the user, agent, application and resulting record change.
Data sovereignty — ues.io runs in containers on the customer’s own infrastructure, in the customer’s own cloud account, or in ues.io’s hosting in Switzerland or the EU.
From Claude-built prototypes to governed applications
Customers who built their applications on Claude, then needed user governance and security to take them into production, can move them onto ues.io, where permission sets, record-level security and audit trails are platform features rather than custom code.
“We built MCP into the application layer, not around it. When an app exposes a tool to an agent, the same permission set and record-level security that apply in the user interface. There is nothing extra to configure and nothing to drift out of sync.”
Mario Richard, Head of Engineering, ues.io
Availability and launch webinar
The Gotthard Release (ues.io 2026.10) is available from 21 October 2026. A free tier is available for evaluation. Production and enterprise tiers add SLA and support. For pricing, contact sales@ues.io.
ues.io will demonstrate the full governed-agent scenario live in a launch webinar on Wednesday 21 October 2026 at 11:00 US Eastern, 16:00 UK and 17:00 Zürich: an agent acts for a named user, reads only that user’s records, proposes a sensitive write, obtains approval, executes it, and produces a trace of the change. Register for the webinar, or see the full webinar series.
About ues.io — ues.io is a Swiss enterprise application platform for building MCP applications and running AI agents under governance. Applications, data, workflows and agents share one permission model, with swappable LLMs and SLMs, a Git-native development lifecycle, and deployment on customer infrastructure, in the customer’s cloud, or hosted in Switzerland or the EU. Built and operated from Zürich since 2020, ues.io serves regulated enterprises in banking, wealth management, insurance, healthcare and the public sector. Learn more at ues.io.
Media contact — Sibylle Baxter, Communications Officer, ues.io GmbH, Zürich, Switzerland, sibylle@ues.io