A model built from your systems
Eraser builds the model from your code, cloud accounts, CMDB and EA tools, and a rescan brings in what changed. In Structurizr, someone writes the model in the DSL; only Java components come from code.
The AI-native Structurizr alternative
C4 models built from your code, cloud and CMDB, drafted and refined with AI, and explored as live maps anyone can open.
Eraser vs Structurizr
How Eraser compares to Structurizr
Structurizr is the reference implementation of C4 as code, built by the model’s creator, Simon Brown: one DSL workspace generates every view, the viewer drills down and animates, and it is open source and runs on your own infrastructure. Eraser builds the model from your code, cloud accounts and CMDB instead of a hand-written DSL, drafts and refines it with AI or on a canvas, and shares it as an interactive page anyone can open.
| Compare | Eraser | Structurizr  |
|---|---|---|
| AI diagramming | Supported: Yes Powered by frontier models: architecture, network, cloud, process and more | Not supported: No No built-in AI; the docs show prompts for outside chat tools |
| Edit with AI | Supported: Yes Describe the change; the diagram updates | Not supported: No Edit the DSL by hand or through an external agent |
| AI follows your standards | Supported: Yes Presets of rules, templates, reference architectures and brand styles | Not supported: No Inspections check the finished model; nothing guides the AI |
| MCP server for AI agents | Supported: Yes AI assistants and coding agents, like Claude, Copilot and others, can create, read and update diagrams | Supported: Yes Open source; validates and exports DSL, and creates or updates workspaces on your server |
| Generate diagrams from live cloud environments |
| Not supported: No Cloud themes style the boxes; nothing reads an account |
| Generate diagrams from CMDBs and EA tools |
| Not supported: No |
| Generate diagrams from codebases |
| Partially supported: Java only A component finder extracts components from compiled Java code |
| Expandable views | Supported: Yes Click into any box to open its own view, then move from a system overview down through layers of detail | Supported: Yes Double-click an element to open the next level of detail |
| Animations | Supported: Yes Animated flows show how a request or data moves step by step, so processes are easier to follow | Supported: Yes Step through animations in static and dynamic views |
| Real-time collaboration | Supported: Yes In the browser, with comments and guests | Not supported: No Teams collaborate through git and pull requests |
| Deployment options |
|
|
| One model, many views | Supported: Yes Boxes stand for model records, reused across views | Supported: Yes Every view is generated from one DSL workspace |
The difference
Why teams choose Eraser over Structurizr
AI built in
Eraser AI drafts C4 maps from a prompt or your data sources, and you refine them by prompt or on the canvas. Structurizr has no built-in AI; its docs show prompts for outside chat tools.
No DSL to learn
Architects, product owners and auditors can change a map on the canvas or by prompt. In Structurizr, every change to the model is a change to DSL code.
Frequently asked questions
-
Does Eraser support the C4 model?
Yes. Eraser draws the C4 model as maps: system context, container, component and code views you click down through, with flows that walk a request across them. Explore a food-delivery platform, the four levels or a monolith carved into services, or see how Eraser compares with other C4 model tools.
-
How do Eraser’s C4 maps stay up to date?
A map is a set of views over a model. When you rescan a data source, such as an AWS account or a ServiceNow instance, its model takes the changes and every map built on it shows them. Eraserbot also watches your GitHub and GitLab repositories and updates diagrams drawn from code as it changes.
-
Can people without an Eraser account view a map?
Yes. Export a map to HTML and you get an interactive page, with its views, drill-down and flows, that anyone can open in a browser. It travels as a file, so it can sit in a SharePoint library or on a Confluence page. Maps also export as images and PDFs.
-
Can Eraser run in our own cloud or on-premises?
Yes. Eraser runs as multi-tenant SaaS, as single-tenant SaaS, as a private cloud in your own cloud, or on-premises, including air-gapped. See security for how to set it up.
-
Can Eraser generate diagrams with AI?
Yes. Eraser AI runs on frontier models. Describe a system, paste code or upload a file, and it drafts an architecture diagram, flowchart, ERD, sequence diagram, BPMN process or almost any other diagram, and you can keep editing it by prompt, as code or on the canvas.
-
Can Eraser draw diagrams from my live infrastructure?
Yes. Connect AWS, Azure or Google Cloud, a CMDB such as ServiceNow, VMware, Cisco networks, a Git repository and other data sources, and Eraser draws architecture diagrams from what you actually run.
-
Does Eraser provide an MCP server?
Yes. The Eraser MCP server lets an agent create, search, read, update and export diagrams from clients like Claude, Cursor, VS Code and ChatGPT. It is free to use, subject to the AI credit limits of your plan. See the MCP client catalog for setup instructions.
-
Can I trust my data with Eraser?
Eraser is committed to enterprise-grade security. We are SOC II type 2 certified and annually penetration tested. Security documentation is available on our trust portal, or reach out to hello@eraser.io. Additionally Eraser will never use user data for training models. Nor will our LLM API providers.
Diagram in seconds with Eraser
Skip the hand-written DSL. Map your architecture from the systems you already run, or generate your first diagram for free.