Your cart is empty
Browse our products and add items to your cart.
Items added to your cart
Your items have been saved.
View and edit your selections by proceeding to your cart.
Industries we empower
Solve & connect
Your cart is empty
Browse our products and add items to your cart.
Items added to your cart
Your items have been saved.
View and edit your selections by proceeding to your cart.
Industries we empower
Solve & connect
Search our site for products, services, training and support
Connect AI agents directly and securely to your Bentley applications.
Bentley’s Model Context Protocol (MCP) servers connect AI agents, including Claude and GitHub CoPilot directly to Bentley applications. Engineers can query design data, automate complex workflows, and drive structural analysis and CAD modeling tasks using plain language. No coding experience required. MCP makes Bentley the AI extension layer for infrastructure engineering.
AI-native structural engineering
Check out these key features of Bentley MCP servers:
The STAAD.Pro MCP server connects AI agents such as Claude directly to the world’s leading structural analysis software. Engineers can automate complex load applications, model revisions, and analysis workflows using plain language, without manual interface interaction or scripting knowledge.
Example: Tell the AI to ‘Apply vertical and horizontal hydrostatic loads to the tank compartments assuming each is full of water’ and STAAD.Pro applies the correct surface loads automatically, preparing the model for stress analysis.
Key capabilities: Natural language model creation and revision | Automated load application | Multi-step structural analysis workflows | Compatible with Claude, GitHub Copilot, Gemini | Secure sandbox – no file system or OS access outside STAAD.Pro.
The MicroStation MCP server connects AI agents directly to your design files, letting infrastructure designers analyze data, automate modeling tasks, and build workflows using plain language. With full access to MicroStation’s Python API, every element, level, reference, and view in your DGN file is accessible to the AI.
Engineers ask natural language questions and the AI connects to MicroStation to extract data or create and modify elements as required. No Python experience needed – prompting skill, not programming skill, is the key competency.
Key capabilities: Full read, write, modify, and create access via Python API | Natural language queries across all DGN file data | Automate repetitive modeling and documentation tasks | Compatible with Claude and GitHub Copilot | Secure environment blocks all file and network access outside the active design.
*Early Access: MicroStation MCP Server is released to a restricted set of users, rather than being generally available to all users.
The iTwin IoT MCP server connects Claude and other AI agents to your live and historical sensor data, giving everyone from field technicians to C-Suite executives a direct, expert-level interface to their sensor data. Teams can use plain conversation to investigate alerts, apply engineering logic, run ML analytics, generate reports, and render custom HTML visualizations.
Example: Ask the AI “Why did the slope inclinometer on Dam 4 trigger an alarm overnight? Pull the last 48 hours of readings and rainfall” to instantly retrieve the telemetry and pinpoint the trigger. Follow up with “Compare this to the baseline, is the slope actively moving?” to run a real-time trend analysis on adjacent sensors and confirm stability.
Key capabilities: Natural language sensor querying | Contextual alert investigation | Custom engineering KPI calculations | Advanced ML-driven trend, correlation, and classification workflows | Live asset hierarchy management and equipment provisioning | Secure enterprise sandbox—no direct OS or file system access outside the iTwin IoT environment.
Power Line Systems’ PLS-CADD MCP server connects AI assistants directly to your PLS-GRID environment, enabling engineers to query live project data, review revision history, identify overloaded structures, investigate terrain coverage, and uncover project insights using natural language.
Embedded within PLS-CADD, the server routes AI requests through the authenticated PLS-GRID connection already established by the application, allowing assistants to work with current, live data. AI assistants operate with the logged-in user’s existing permissions and provide read-only access, enabling secure analysis and reporting without the ability to create, modify, or delete data.
Key capabilities: Query live PLS-GRID data across projects, structures, spans, and sections | Access complete revision history and change tracking | Identify overloaded structures and terrain, LiDAR, or vegetation management gaps | Audit user and ACL permissions | Download specific file revisions for review and archival | Read-only access protects server data | Compatible with GitHub Copilot, Claude, Gemini, and other MCP clients | Secure local deployment with no firewall changes required.
Power Line Systems’ TOWER and PLS-POLE MCP servers connect AI assistants directly to the structure model open in the running application, enabling engineers to analyze structures, review results, compare models, validate designs, and generate deliverables using natural language.
The MCP servers are embedded directly within TOWER and PLS-POLE, enabling AI assistants to interact with live models through the same command API used by PLS-CADD and third-party automation tools. No external server or network connection is required. The MCP server starts automatically with the application and shuts down when it closes, providing a seamless and secure AI-assisted workflow.
Key capabilities: Full analysis, model check, and export capabilities via application command API | Natural language queries across open structure models | Automate allowable span generation and file comparisons | Compatible with GitHub Copilot, Claude, Gemini, and any MCP-enabled client | Secure loopback transport restricts access to AI clients running on the local machine.
“The reality of the infrastructure sector is that there is no room for approximation, let alone hallucination. A model that is 90% right is a useful start; a structural analysis that is less than 100% right is a catastrophic liability.”
Looking for additional MCP resources?
MCP stands for Model Context Protocol, an open standard that allows AI agents to connect to and interact with software applications. Bentley’s MCP servers expose the full Python APIs of Bentley applications so that AI agents can query, create, and modify engineering data in real time. This means AI agents like Claude can drive complex structural analysis or CAD workflows on behalf of the engineer, using plain language instructions.
No. MCP servers translate plain-language instructions into Python API calls automatically. Engineers need no programming or scripting knowledge. The more specific and contextual the prompt, the better the output, so familiarity with effective AI prompting (describing intended outcomes, referencing specific elements or standards) improves results. Bentley will provide guidance on effective prompting practices.
Bentley’s MCP servers implement the open MCP standard and are compatible with any MCP-compatible AI agent, including Anthropic’s Claude, GitHub Copilot, and Google Gemini. Users are not locked into a single AI platform. They can choose the AI agent their organization already uses or evaluates, and connect it to Bentley applications through the same MCP interface.
Yes, your data is secure, with important distinctions in how different Bentley MCP server environments manage access, particularly concerning filesystem and network operations.
The STAAD.Pro MCP server runs in a secure sandbox. It has no access to the file system or operating system outside of STAAD.Pro, and its scope is intentionally limited to STAAD.Pro operations.
For the MicroStation MCP server, the Python Code Execution environment is secure, in that it blocks access to the internet and filesystem. However, it’s important to understand that it may be possible for one of MicroStation’s ‘key-ins’ (commands) executed via MCP to access the file system. For example, a key-in such as “File Fence” with a specified path and filename could perform its expected work and create a .dgn file. This means that while the direct scripting capabilities of the Python environment itself are sandboxed from the filesystem and network, these environments can trigger existing functionalities within the host application (MicroStation) that inherently possess such access.
Bentley Copilot, the in-application AI assistant, uses the MCP server as its primary interface to interact with Bentley application APIs. By publishing this as an open MCP server rather than a private integration, Bentley gives users the choice: use Bentley Copilot inside the application, or connect your own AI agent (Claude, GitHub Copilot) to the same capabilities through the open standard.
MicroStation and STAAD.Pro are the first in a planned rollout of MCP servers across the full Bentley portfolio. As more products add MCP servers, users will be able to orchestrate end-to-end workflows across multiple Bentley products, and even combine Bentley MCP servers with those of other vendors. All driven by a single AI agent.
