Align
We began by deeply understanding how the Pharma Company’s Legal team managed contracts day to day.
Our UX and Product teams mapped every step of the process with the Legal Manager. They analyzed how contracts were located in SharePoint and how metadata was extracted and compared manually. This helped uncover the main bottlenecks: tedious searches, duplicated effort, and an overreliance on external attorneys.
These insights revealed a clear opportunity to bring AI into the process responsibly, without disrupting daily workflows.
Prioritize
From our discovery work, we identified four daily tasks where AI could make the greatest impact: searching, reading, and comparing contracts.
These became the foundation of our Proof of Value:
Contract content discovery: ask natural-language questions about contract data.
Metadata extraction: automate how key contract fields are identified and stored.
Contract search: combine keyword and metadata search for faster, smarter retrieval.
Template comparison: highlight differences between a contract and its standard template.
Together, these use cases defined the PoV scope — proving how AI could reshape legal operations at scale.
95–100%
search accuracy
through hybrid AI agents merging metadata and keyword queries
Manual effort in contract search and comparison
eliminated through automated data extraction and AI summaries
Prove
We built an AI-powered platform that mirrored the familiar SharePoint interface but delivered entirely new capabilities behind the scenes.
AI-driven contract search
Implemented a dual-mode AI Agent with metadata-based (SQL) and keyword-based (Azure AI Search) tools.
Results: 95–100 % keyword-search accuracy and 71 % metadata-search accuracy.Metadata extraction automation
Deployed multimodal LLMs to parse contracts and auto-populate fields such as expiration or signature dates, achieving high reliability with exceptions flagged for legal validation.Contract chatbot (“Madra”)
Enabled users to query entire contracts conversationally without RAG pipelines, reaching ≈ 84 % answer accuracy.
Template comparison Designed AI-generated summaries highlighting key deviations, replacing manual side-by-side reviews.
Integrate
We integrated these AI components into a cohesive UX platform resembling the Legal team’s existing tools to drive adoption.Natural-language filters, guided prompts, and AI-generated summaries reduced friction and saved time, while preserving data governance within Pharma Company’s Azure and Databricks stack.Agile sprints and proof spikes allowed the team to iterate fast, validate AI behavior, and refine workflows in real time.
Team capacity
eliminated through automated data extraction and AI summaries
Scale
Beyond the PoV, we designed a roadmap for growth toward a production-ready CLM platform:
Cost-Benefit Analysis – Azure migration plan and cost projections.
Feature Expansion – future AI modules for clause automation, risk scoring, and compliance tracking.
User-Centric Refinement – prioritized improvements to handle metadata edge cases and ensure adoption.
