Discover how the open‑source Agentic Compliance Insights module brings generative AI into SitecoreAI to audit and improve your pages.
This post covers the motivation for automating compliance checks, the multi‑agent architecture behind the module, setup instructions, and how you can customize the solution for your own organization.
Why content compliance matters
Marketers and content editors publish a tremendous amount of copy every day – from promotional headlines to product pages, blog posts and regulatory notices.
Manual review can miss mistakes: biased language, undisclosed personal information or incorrect statistics.
Regulatory frameworks like GDPR, CCPA and industry‑specific policies add another layer of complexity.
In a world where content velocity is only increasing, relying solely on human judgment is risky and costly.
Introducing Agentic Compliance Insights
Agentic Compliance Insights is an open-source Sitecore Marketplace module that integrates directly into the SitecoreAI Pages Editor.
When editors open a page, the module analyzes the content in the background and flags issues in a side‑panel.
This starter kit uses a collection of AI agents to check for compliance violations, factual inaccuracies, missing citations, biased phrasing, personally identifiable information and even potential plagiarism.
Because the module runs as a page context panel extension, it inherits the editor’s authentication and does not require storing API keys in the browser.
Multi‑agent architecture
The solution comprises two main components: a Next.js frontend that renders inside the SitecoreAI editor built leveraging new Sitecore Marketplace SDK Starter kit, and a Python backend built on FastAPI and the Microsoft Agent Framework.
The backend orchestrates multiple specialized agents:
| Agent | Purpose |
|---|---|
| Compliance agent | Checks content against configurable rules defined in an Excel sheet |
| Fact‑checker | Extracts factual claims and verifies them against trusted sources |
| Citation agent | Detects and validates inline citations |
| Bias detection agent | Highlights exclusionary or biased language |
| PII agent | Identifies emails, phone numbers and other personal data |
| Plagiarism agent | Flags potential plagiarism |
| Fix application agent | Applies recommended changes automatically |
These agents execute concurrently, managed by a supervisor in the backend, and their findings are aggregated into an easy‑to‑read audit report.
Editors can view the checklist of violations, read explanations and apply suggested fixes with a single click.
The module also provides a REST API, asynchronous processing and multi‑agent orchestration for broader integrations.
Sitecore Marketplace integration
The frontend is built as a Page Context Panel extension and communicates with SitecoreAI via the Marketplace SDK.
Because authentication is handled by SitecoreAI, the extension does not require direct API keys from editors, making it safe to deploy in enterprise environments.
Technology stack
The project uses modern, open‑source tools:
- Frontend: Next.js 15 (app router), Chakra UI and the Sitecore Marketplace SDK.
- Backend: Python 3.11, FastAPI, and the Microsoft Agent Framework.
- Data processing: OpenPyXL for Excel parsing and Azure OpenAI (or OpenAI API) to run large‑language models.
This modular architecture makes it straightforward to extend or swap components—whether you want to use a different UI library, host the backend on your own infrastructure or integrate additional agents.
Setting up the module
Prerequisites
To run Agentic Compliance Insights locally, you’ll need:
- Node.js 18+ and npm
- Python 3.11+
- An Azure OpenAI deployment or an OpenAI API key
- A SitecoreAI instance with Marketplace access
Sitecore Marketplace setup
- Create the extension – log in to your SitecoreAI instance, open App Studio and create a new app called Agentic Compliance Insights.
Choose the Page Context Panel extension type and set the URL tohttp://localhost:3000during development. - Start the apps – in one terminal, run the Next.js frontend (
npm run dev), and in another, run the FastAPI backend (python -m uvicorn src.app:app --reload --port 3001). - Navigate to the editor – open any page in SitecoreAI’s Pages Editor.
You should see the Agentic Compliance Insights icon in the toolbar; clicking it opens the panel with real‑time analysis.
Configuration
Configuration is handled via environment variables.
In the backend .env file, set your Azure OpenAI endpoint, deployment name and API key.
You can also point to a custom compliance rules Excel file and supply a WEB_LLM_API_KEY if you need to perform web searches, especially for fact and citation validation agents.
On the frontend, specify the backend base URL (default http://localhost:3001).
Customizing compliance rules and fixes
One of the strengths of this module is its configurability.
Compliance rules are stored in an Excel sheet and can be tailored to your organization’s policies.
Each rule defines a category, description, severity and example—allowing you to flag anything from accessibility violations to corporate style guidelines.
You can also modify the prompts used by each agent, adjust AI models, implement role‑based access control or extend the UI to include richer reporting.
Agentic Compliance Insights in action
Below is a demo of the module running inside the SitecoreAI editor:
Conclusion and next steps
Automating content governance with AI doesn’t remove the need for human judgment—but it does free up your team to focus on creative storytelling instead of repetitive audits.
By plugging Agentic Compliance Insights into SitecoreAI, you get an always‑available assistant that flags issues early and suggests actionable fixes.
Because the project is open source under the MIT license, you can adapt it to your workflows, swap out models or contribute enhancements.
If you’re interested in trying it out, clone the repository from GitHub, follow the setup instructions and start analyzing your pages today.
Feedback, issues and pull requests are welcome—this module is for the community, and your contributions will help improve it for everyone.