AI-Powered Document Summaries: Delivering Instant Value to Legal Professionals
For Better Regulation, we implemented an AI-powered solution that automatically summarizes complex legal documents and identifies key obligations within them. Built on Drupal and AI Automators, the system replaces time-consuming manual analysis with clear, structured insights, making regulatory content easier to understand, manage, and reuse at scale.
Key results
3 summary types generated
Long overview, short abstract, and obligations extraction.
2-350 pages summarized
Instant insights from documents of any length.
Automatic processing
Background generation requires no editor action.
Obligations extraction
Critical compliance requirements identified automatically.
Phased user rollout
Opt-in approach building trust with subscribers.
Measurable time savings for users
30s relevance check instead of minutes or hours of manual reading.
About Better Regulation
Challenge: information overload for legal professionals
The user problem: too many documents, too little time
Legal professionals subscribing to Better Regulation face a fundamental challenge: they need to stay current with legislation, but they don't have time to read everything.
The scale of the problem:
- hundreds of new and updated documents published monthly,
- documents ranging from 2 to 350 pages each,
- multiple jurisdictions to monitor (UK, Ireland, EU),
- frequent legislative amendments and updates,
- time-critical information buried in lengthy statutes.
User pain points
What users needed
The documents themselves presented additional complexities.
Solution: three types of AI-generated summaries
Automatic value addition to every document
Droptica worked with Better Regulation to implement an AI-powered summary system that automatically generates three complementary summary types for every document on the platform.
The solution runs in the background, requiring no editor action, and provides immediate value to users viewing any document.
1. Long summary (~200 characters)
Purpose: comprehensive document overview.
A detailed summary that captures:
- the document's main purpose and scope,
- key themes and provisions,
- context for why the document exists,
- who it affects and how.
Example from actual implementation:
The Bankruptcy Act of 1986 serves as a critical piece of legislation aimed at addressing insolvency and establishing frameworks for bankruptcy proceedings. This act not only sets forth the legal parameters for individuals and businesses in financial distress, but also lays the groundwork for debt resolution. It's essential for legal practitioners to comprehend the intricacies of this act, especially when advising clients on financial recovery or navigating insolvency scenarios.
User benefit: quickly understand whether the document is relevant to their work without opening the full text.
2. Short summary (single paragraph)
Purpose: concise abstract for quick scanning.
A brief, focused summary that:
- highlights the most critical information,
- provides just enough context for decision-making,
- enables rapid scanning of multiple documents,
- helps users prioritize what to read in detail.
User benefit: scan dozens of documents quickly to identify which require detailed reading.
3. Obligations summary (⭐ key differentatior)
Purpose: automatic extraction of legal obligations and requirements
This is the most innovative and valuable summary type. It automatically identifies:
- legal obligations and requirements within the document,
- compliance points that professionals must be aware of,
- relevant license types or affected parties,
- key regulatory requirements extracted from complex text.
User benefit: compliance officers and legal advisors can immediately identify if a document contains obligations relevant to their organization or clients without reading hundreds of pages.
Visual example: AI summary in action
Key elements visible in the interface:
- AI Summary sidebar accessible while viewing the document.
- "AI text summary" and "AI extracted obligations" tabs for different summary types.
- Export options - users can email or download PDF.
- Professional presentation - summaries formatted for easy reading.
- Always available - no additional clicks or waiting required.
How it works: background processing
Technical implementation
Robust architecture
The summary generation uses the same proven technology stack as the document categorization feature:
- Drupal 11 - content management and user interface,
- Drupal Automators - workflow orchestration and prompt management,
- Unstructured.io - clean PDF text extraction,
- GPT - AI analysis and summary generation,
- RabbitMQ - background job queue,
- Watchdog - comprehensive logging.
Results: delivering user value and product differentiation
User value: time savings and better coverage
While detailed usage metrics are still being collected during the phased rollout, early feedback indicates significant user value.
Specific use cases from early feedback
Technical performance
Quality assurance process
Human review layer: while summaries are generated automatically, quality is maintained through:
- Sample review: editorial team regularly reviews summary quality.
- User feedback: opt-in users can report issues or inaccuracies.
- Automated checks: validation that summaries meet length and format requirements.
- Continuous monitoring tracks metrics like summary length, processing time, user engagement.
Iterative improvement:
- Collect examples of particularly good summaries.
- Identify patterns in user corrections or feedback.
- Refine prompts based on lessons learned.
- A/B test prompt variations.
- Update as AI models improve.
Integration with Drupal
The technical integration showcases Drupal's flexibility.