ABOUT
Ai Bloks (Alpha):
Designing AI for Legal Professionals
Designed the alpha version of AI Bloks now LLMware, an AI-powered platform for legal and enterprise document analysis. I translated raw prototypes into intuitive UX/UI, built the early visual system, and helped onboard first users—while learning AI fundamentals hands-on during the process.
Project Background & My Role
Drawn by my interest in AI, I worked with AI Bloks during its earliest phase to help shape the alpha version of their document intelligence platform. The product was still evolving, with features changing frequently and no real UI in place. I was the only UX designer on the team, working with early, functional prototypes from engineers. My focus was to make the technology accessible—refining user flows, building visual consistency, and creating a clean interface users could actually navigate.
The Challenge
The Alpha Product Reality
I encountered a fascinating but daunting challenge: transforming a functional but fragmented AI prototype into a coherent, user-friendly alpha product. The engineering team had built powerful AI capabilities, but the interface was a collection of disconnected screens with broken UX patterns and inconsistent design elements.
My Learning Curve
This was my first AI project, and I had to learn about AI—ML, NLP, and LLMs—in real time while designing for it. With the tech evolving rapidly and few UX references available, I was learning and building at once.
The Evolving Product Challenge
The real challenge wasn’t just the complex tech—it was keeping up with a constantly evolving product. Features changed weekly, and I had to quickly translate raw, functional prototypes from engineers into cohesive, user-friendly experiences.
Core Problems to Solve
Broken User Journey
Disconnected screens with no logical flow between AI-powered features
Technical Complexity Exposed
Raw AI outputs and processes were visible to end users
Inconsistent Interface
Each feature felt like a separate tool rather than part of a cohesive platform
Rapid Feature Evolution
Interface designs needed to be flexible enough to accommodate changing AI capabilities
Our mission was to solve real pain points for legal and enterprise teams struggling with
High costs and time drain of manually reviewing documents
Messy or incomplete searches using traditional keyword systems
Data locked away in PDFs, PowerPoints, scanned files, and outdated formats
Manual tagging of data for AI model training
Poor visibility into contractual obligations and risks
Users
We were building for a cross-functional audience
Legal teams and law firms drowning in contracts
AI scientists preparing data for NLP models
In-house analysts who validate AI-extracted results
Business users with no technical or AI background
My Design Process & Learning Journey
Working with Engineering Prototypes
My typical workflow involved receiving technical briefings from the engineering team about new AI capabilities they had developed. They would demonstrate functional prototypes - often impressive in their AI capabilities but with minimal user interface consideration. My role was to:
Understand the Technology: Grasp what the AI was actually doing behind the scenes
Identify User Value: Determine how this capability would benefit legal professionals
Design User Flows: Create intuitive pathways for users to access and understand AI outputs
Collaborate with UI Developer: Work closely with our UI developer to implement cohesive, user-friendly interfaces
Collaboration with Leadership
Working with CEO Namee and CTO Darren was invaluable. Their extensive experience in both business and technology provided crucial context for design decisions. They helped me understand:
The strategic vision for each AI feature
Technical constraints and possibilities
Market expectations and competitive landscape
Long-term product evolution plans
My UX & UI Contribution
We were building for a cross-functional audience
Redesigned early, broken UI prototypes into cohesive user journeys
Created a modular design system that scaled across features
Designed key flows: search, document view, analyst review, template setup, export
Worked with developers to refine microinteractions and onboarding
Crafted the first version of the platform's visual identity—type, color, icons, layout
Simplified dense AI workflows so even non-technical users could start using the platform in minutes
Some key interfaces I designed
*Due to confidentiality agreements, I'm unable to share detailed screenshots or proprietary information
The platform can process and search through a wide variety of file types, including PDF, Word, Excel, PowerPoint, HTML, Outlook, G Suite, Dropbox, and CSV.... It can even find embedded images and tables within these documents
No-Tag Data Sets: The platform automatically identifies linkages and statistical correlations within documents to create ML/NLP data sets without requiring manual tagging. This allows models to be trained on an organisation's specific language and data automatically
Customised AI Models: Users can train and deploy ML and NLP algorithms using their own data to create customised AI models that "speak your language". This can be done quickly, in less than a day, and does not require users to be AI experts or have coding skills
Ai Bloks provides an intuitive, Google-like search experience across all your documents, allowing for natural language queries. This enables users to find content based on the intent of their search, rather than needing exact words or phrases.
Contract AI is a specific product offered by Ai Bloks that features a special purpose LLM called "richiept," which is specifically trained on contracts and agreements.
Key Value Extraction: It includes an inference template builder and pre-built process flows to extract key values and provisions from contracts, such as yes/no selections, multiple-choice answers, or text.
User Interface: It provides a user interface for analysts to confirm or correct extracted answers, and results can be exported as a CSV file.
Deployment: Contract AI can be deployed in a secure, private cloud environment using Kubernetes, ensuring no data leaves the client's control. It is described as a TRL (Technology Readiness Level) 9 system, proven through successful mission operations.
Alpha Product Success Metrics
Coherent User Experience: Transformed disconnected AI prototypes into a unified, intuitive platform
Rapid Feature Integration: Reduced time to implement new AI capabilities from weeks to days
Stakeholder Approval: Successfully demonstrated product vision to investors and early customers
Foundation for Growth: Established design patterns and component library for future development
Contract AI Testimonials
"I asked mt team how long would it take to do this analysis, they said 1 week... AI Bloks did it in 6 minutes"
Phil Helms, Principal
"We relied on cutting edge Ai Bloks artificial intelligence software to greatly reduce attorney time by searching a reporting on our entire month-ling trial"
Dean Orgin, Principal
"We are excited about Ai Bloks' generative AI solutions. Their specialized AI models and pre-built process modules for text intensive workflows make adopting AI practical, secure and cost effective. "
Chris Wilkinson, SVP
Reflection
I chose this project because of my growing interest in AI and wanted to explore how design could make it more accessible. It was my first AI-focused product—and a hands-on crash course in turning complex ideas into simple, usable interfaces.
Since then, the product has evolved into LLMWare, a platform that helps enterprises build and deploy private LLMs securely. I'm proud to have shaped its early UX and visual identity.
Proud to be a part of a female-led organization with an extended team of 70% women and strongly
promote inclusivity in the workplace.
Thanks for reading.
"Aditi brings excellent professionalism and attention to detail to her work, and she works very collaboratively and is responsive to her clients' needs. I highly recommend Aditi for any UX design work!"
Namee Oberst (Co-Founder at LLMWare)
Sitemap
Social