On August 19, 2026, UiPath (NYSE: PATH) launched Maestro Flow, a developer-first orchestration canvas that connects AI coding agents directly to governed enterprise production. The announcement targets what UiPath’s Chief Product and Technology Officer Raghu Malpani identified as the real enterprise AI bottleneck: “Enterprises don’t have an agent problem; they have an orchestration problem.”
Maestro Flow is available in public preview immediately. It supports Claude Code, Cursor, GitHub Copilot, OpenAI Codex, Google Gemini CLI, and Windsurf as first-class authoring tools, and it runs on the same Maestro orchestration engine that powers enterprise automation at global scale.
The Problem: Agents Build Fast, Production Ships Slow
Every enterprise team building with AI coding agents has hit the same constraint. A coding agent can scaffold a working automation in an afternoon. Getting that automation into governed enterprise production, with audit trails, RBAC, observability, and integration with existing systems, takes weeks. Sometimes it never ships at all.
The prototype and the production artifact are different things. They live in different environments, get reviewed by different teams, and require different skill sets to maintain. When something fails in production, there is often no shared artifact that connects what was built to what is running. Process logic fragments across notebooks, scripts, and ad-hoc integrations.
According to UiPath’s announcement, without a critical orchestration layer, “process logic becomes fragmented, difficult to govern, and costly to maintain. Prototypes cannot ship, work stalls while waiting on review, automations are not tracked, and the cost of experimentation itself becomes a brake on progress.”
The economics of this gap are significant. Research from IT consulting firm Triple Minds, cited in industry coverage, estimates that building a production-ready agent workflow, accounting for context management, tool governance, and orchestration overhead, can add up to more than $100,000 per deployment. Teams that pay that cost repeatedly, for each prototype that fails to ship, stop experimenting.
What Maestro Flow Is
Maestro Flow treats the prototype and the production deployment as the same artifact. A developer builds a Maestro flow on a canvas or via CLI. That flow runs in production on UiPath’s Maestro execution engine, with enterprise governance, fault-tolerant runtime, and execution tracing included. No re-platforming. No handoff. No separate build for production.
The core architectural decision is that coding agents are first-class participants in authoring, not just in code generation. UiPath provides platform skills as MCP (Model Context Protocol) tools that give any supported coding agent knowledge of Maestro’s conventions, schemas, and validation rules. A developer can describe an automation in natural language inside VS Code or Cursor. The coding agent uses the platform skills to generate a Maestro flow that meets UiPath’s production standards, runs local validation checks, and publishes to the cloud through a single command. The artifact the agent wrote is the one that ships.
That is a different model from most enterprise orchestration. Typically, AI coding agents generate code that requires a human to review, adapt, and re-platform before it can enter a governance-controlled production environment. Maestro Flow’s MCP integration removes that gap by giving the coding agent domain-specific knowledge of the production environment from the start.
The Supported Coding Agent Ecosystem
Maestro Flow is designed to work with any coding agent currently in enterprise use:
| Coding Agent | Integration | Status |
|---|---|---|
| Claude Code | MCP tools and CLI | Available now |
| Cursor | MCP tools and VS Code extension | Available now |
| GitHub Copilot | VS Code extension | Available now |
| OpenAI Codex | MCP tools and CLI | Available now |
| Google Gemini CLI | MCP tools | Available now |
| Windsurf | VS Code-compatible extension | Available now |
The agent-agnostic design reflects enterprise reality. Platform teams rarely standardize on a single coding agent. Teams using Claude Code work alongside teams using Cursor, and both need their automations to enter the same governed production environment. Maestro Flow provides that shared production layer without requiring a vendor choice.
F5’s 2026 State of Application Strategy Report found that 77% of organizations now say inference, rather than training or fine-tuning, is their dominant AI activity, and that the average enterprise manages seven AI models. As inference volume scales and organizations run more agents across more systems, the orchestration and governance layer becomes the strategic asset, not the model itself.
Three Ways to Author a Maestro Flow
Studio Web is a browser-based canvas available with no installation. Teams can collaborate in real time and use AI-assisted design to iterate on flows. It is the fastest path from idea to running process for teams that want to prototype without environment setup.
VS Code Extension brings Maestro Flow authoring into the IDE where most developers already work. It supports live validation against UiPath’s production standards, so developers get feedback on structural issues before they publish. Cursor, Windsurf, and other VS Code-compatible environments work through the same extension.
CLI and MCP Integration exposes the full Maestro Flow build lifecycle to coding agents. A developer opens a terminal, initializes a project, and instructs their coding agent to build a specific automation. The agent uses UiPath’s platform skills to generate, validate, and publish in a single workflow. No browser, no canvas, no context switching: the coding agent builds directly into production.
This authoring flexibility addresses a common enterprise failure mode: governance tools that require developers to leave their preferred environment create friction that leads to shadow deployments and ungoverned processes. Maestro Flow’s multi-interface approach lets enterprise platform teams enforce governance without mandating new tooling.
Why This Lands Now
The enterprise AI market has spent two years watching coding agents improve dramatically. Agents now routinely write production-quality code. The constraint that remains is not what agents can build; it is whether what they build can enter an enterprise’s governed production environment.
Cursor’s launch of Origin last week illustrated the same tension from a different angle: a coding agent vendor building code hosting to control the full development lifecycle. Maestro Flow represents the inverse: an enterprise orchestration vendor building coding agent integration so that what gets built natively can ship natively.
Both moves reflect the same underlying pressure. The gap between “an agent can build this” and “this is running in production with an audit trail” is where enterprise AI ROI actually lives. Addressing it requires either the coding agent vendor to extend into enterprise operations, or the enterprise operations vendor to extend into coding agent tooling. UiPath has chosen the latter, and the MCP integration model gives it a path that does not require developers to switch agents.
For enterprise GTM teams and operations builders, the practical implication is concrete. Automations that coordinate multiple data sources, agent reasoning steps, human approvals, and system integrations, the kind of workflows that drive real business process change, can now be built at coding agent speed and deployed into an enterprise-grade runtime without a separate engineering engagement to make the handoff work.
Governance Built In, Not Bolted On
Maestro Flow processes enter UiPath’s existing governance environment automatically. Every process gets execution tracing, instance management, and RBAC. Administrators apply policy through Automation Ops. Identity and audit trails run through Automation Cloud. Compliance and observability come with the runtime, not as a configuration layer developers have to add after the fact.
The Salesforce Agentic Enterprise Index 2026 documented widespread gaps in enterprise AI governance, particularly around agent observability and audit trails. Maestro Flow addresses those gaps structurally by running every coding agent-built flow on a runtime where observability is a default, not an option.
For regulated industries, the implications are significant. A healthcare team building an agentic patient intake process, or a financial services team building an automated compliance workflow, can now use Claude Code or Cursor to build the process and deploy into an environment that already meets their audit and access control requirements. The pattern of orchestration platforms absorbing governance complexity continues to emerge as the practical path for enterprises that cannot staff a dedicated forward-deployed engineering engagement for each production deployment.
What to Do With This
Maestro Flow is in public preview. Developers can try it at uipath.com/product/maestro/flow starting today. For teams using Claude Code or Cursor to automate enterprise workflows, the MCP integration is the immediate entry point: install the platform skills, describe the automation, and let the coding agent build a flow that ships to production rather than to a staging environment that never graduates.
For platform teams evaluating governance architecture, the key question is whether your current toolchain can trace what a coding agent built all the way through to what is running in production. If the answer is no, that traceability gap is where business risk accumulates. Maestro Flow, running on UiPath’s existing enterprise runtime, closes it without requiring a new infrastructure investment.
Maestro Lite, announced alongside Maestro Flow, provides a lighter orchestration option for teams with lower governance requirements. It scales the depth of enterprise management to match the cost and criticality of the work, giving teams a lower-friction starting point that can grow into the full Maestro platform as their agentic deployment complexity increases.