Progress Agentic RAG Enables Trusted, Traceable Answers for a Leading Law Firm's AI Search Experiences
Industries:
Professional ServicesProducts:
Agentic RAGChallenge
A European law firm serving thousands of clients struggled with time-intensive research for complex legal/accounting questions.
Strict EU regulations, such as GDPR, required full AI explainability and traceability. Internal AI builds and vendor evaluations failed to deliver the trusted accuracy and source visibility lawyers needed.
Solution
- Consolidated legal and accounting knowledge within the Progress Agentic RAG solution to power custom AI legal research workflows with built-in governance, traceable answers and full auditability to verify sources.
- Designed tailored retrieval strategies reflecting how lawyers search/interpret evolving legislation, ingest internal knowledge and curate external sources.
- Applied REMi (RAG Evaluation Metrics intelligence) to continuously measure and improve answer accuracy, relevance and reliability over time.
Result
- Transformed client inquiry responses for ~300 professionals with traceable, cited draft answers, enabling rapid validation vs. starting from scratch.
- Processed thousands of questions monthly, cutting manual research time so lawyers could focus on analysis/strategy while serving the growing demand of their clients.
- Launched monetized, client-facing AI legal assistant via secure portal, creating differentiated revenue while putting control into lawyers’ hands for final advice.
Full Story
Challenge
A prominent European law firm serving thousands of corporate and SMB clients relied strongly on specialized legal and accounting expertise to answer complex, regulation-heavy questions. Traditional research workflows were time-consuming, requiring lawyers to search across large volumes of legislation, guidance and internal know‑how before providing an answer. At the same time, the firm needed to uphold strict European regulatory standards, such as GDPR, around confidentiality, explainability and responsibility when using AI in a legal context.
The firm initially explored building its own AI-powered research solution and evaluated multiple vendors, including hyperscalers and other retrieval-augmented generation (RAG) offerings. However, these attempts could not deliver the combination of accuracy, traceability and legal-grade reliability the firm required.
They also lacked end-to-end visibility into how answers were generated, provided limited control over retrieval behavior and made it difficult to demonstrate why a specific clause or legal interpretation was suggested.
The firm’s CTO and Head of AI wanted a platform that could model the way legal professionals research and interpret evolving legislation. They needed precise control over how documents were ingested, chunked, retrieved and presented, plus the ability to track how laws change over time and how those changes impact individual clients. Above all, they required a solution their lawyers could trust, with every answer backed by clear links to authoritative legal sources.
The Progress team understands how the legal industry is required to use technology per European regulations. The knowledge layer that’s built into the Progress solution is continuously being optimized, and the impact we generate by being able to answer questions faster directly improves the way our customers work.
Chief Technology Officer & Head of AI
Law Firm
Solution
The firm selected the Progress Agentic RAG solution as the foundation for an AI-powered legal assistant that could be tailored to its specific workflows. Working closely with the CTO and development team, Progress helped design a retrieval strategy that reflected how lawyers search, interpret and apply legislation. The solution ingests the firm’s internal legal and accounting knowledge, along with carefully curated external content, to provide answers that are both comprehensive and up to date. To support rigorous testing and safe rollout, the firm implemented multiple knowledge boxes for staging and production.
Different retrieval strategies were then employed for distinct use cases, such as employment law, tax topics or specific regulatory domains. This structure lets the team validate changes in a controlled environment before promoting them to production, while keeping data sets cleanly separated by purpose and sensitivity.
Using the platform's model-agnostic experimentation environment (e.g., RAG Lab & Prompt Lab), the firm’s AI team continuously tested different prompt patterns, retrieval configurations and large language models (LLMs).
After comparing models and prioritizing legal accuracy and alignment with regulatory expectations over time, they identified the appropriate LLM and retrieval configuration that delivered the most precise and consistent results for their domain, while preserving the flexibility to switch models as technology evolves.
Evaluation and auditing capabilities (internally referred to as RAG metrics and a monitoring layer) provided detailed insight into system performance.
The team continues to track quality metrics for answers, analyze representative queries and inspect retrieved sources to understand why specific passages were selected. Each response is then delivered with clear citations back to the original legal documents, giving lawyers immediate visibility into the provenance of the information. This feedback loop enables ongoing tuning of retrieval rules, prompts and model selection, fortifying the assistant to continue meeting the firm’s high bar for legal reliability.
Result
With the Progress Agentic RAG solution, the firm has transformed the way its ~300 legal and accounting professionals respond to client inquiries. Lawyers now receive draft answers grounded in the firm's own knowledge and authoritative legal sources, accompanied by clear citations and links to underlying documents. This transparency gives them confidence to validate, refine and finalize responses quickly, rather than starting each research task from scratch each time.
The firm now handles thousands of legal and accounting questions per month through the AI assistant, significantly reducing the time spent on manual research and document review. Lawyers can instead focus on nuanced analysis and client strategy while the system manages the heavy lifting of retrieval and synthesis. This has improved overall productivity and allowed the firm to serve growing demand without compromising quality.
Beyond internal efficiency, the firm has built and launched a customer-facing AI legal assistant powered by the same platform. Select clients can access a secure portal to pose routine legal questions and receive instant, well-grounded answers, backed by the firm's expertise. This new offering creates a differentiated service model and opens additional revenue opportunities, while still putting control into lawyers’ hands for final advice.
The firm continues to benefit from an optimized cost profile for its AI operations. By testing multiple models and fine-tuning retrieval behavior, the team has found an effective balance between token usage and answer quality. They can easily test and swap in newer or more cost-effective models as they become available, maintaining high accuracy while managing spend.