Wanna jump straight to the visuals? I’ve got you
Scope - Four month internship project at the Israel Regulatory Authority, where our team built an AI-powered regulatory database from scratch to centralize all government regulation in Israel. The platform was designed to make regulatory information accessible, clear, and intuitive through advanced search and smart filtering.
Role - Head of Internship Team, UX Research, Wireframing & Prototyping
Tools - Figma, Excel, UXtweak & Bolt
Problem Statement
As no centralized regulatory database has ever existed in Israel, the project aimed to define the foundation for a first of its kind government platform.
Smart Search Engine - Developing an API based search system capable of understanding diverse query types, including misspellings and variations in phrasing.
Real-Time Transparency - Providing up to date regulatory information with change notifications and access to previous document versions.
Process-Oriented Navigation - Presenting search results and document relations in a structured, contextual flow—helping users understand each regulation’s background and dependencies.
Public Awareness - Introducing the Israel Regulatory Authority and its purpose, clarifying its role in unifying, managing, and simplifying access to all national regulation.
KPI'S
As the Regulatory Database is still in its planning phase, success metrics remain theoretical. Three primary indicators were defined to guide future evaluation:
Frequency of Use — Measures how often users are expected to access the system annually (Approximately 10–20).
User Retention — Indicates the percentage of users returning to the platform over time.
Perceived Credibility & Authority — A qualitative measure of trust in the platform as an official and reliable information source. Our target is to Achieve credibility comparable to existing legal and regulatory information systems
Conducting the Research
The research phase focused on understanding user needs and mapping the regulatory ecosystem. It included target audience analysis, user interview analysis, competitor and comparative research, task analysis, card sorting and How might we framing conducted during the wireframing stage.
Target Audience Analysis
To understand who the system should serve, we mapped and analyzed the ecosystem of users interacting with regulatory information. This process identified four main audience groups—each with distinct goals, challenges, and levels of expertise.
Audience Groups
Business Owners – Seek to quickly understand relevant regulatory requirements, avoid fines, and access all forms and procedures in one place. They struggle with complex terminology, unclear processes, and uncertainty about document validity.
Legal Professionals – Need an efficient, centralized database to locate official and updated versions of laws, track amendment history, and save time on cross-checking sources. Their main pain points include scattered data, unclear document status, and time consuming searches.
Regulators – Require coordination tools and a unified view of regulations to maintain consistency and identify overlaps across ministries.
Academia & Civil Organizations – Need open access to reliable data and analytical tools for research, reports, and policy recommendations.
Following a discussion with the product manager, we concluded that the system should be viewed through two core user segments: professional users, who engage with regulation as part of their daily work ,and non-professional users, who interact with regulation occasionally.
User Interview Analysis
To better understand user needs and behaviors, we analyzed around 70 in-depth interviews previously conducted by the Israel Regulatory Authority. This explanatory study aimed to identify what users from different backgrounds — regulators, lawyers, business owners, and citizens — need from a centralized regulatory system and what difficulties they face today. The interviews were reviewed systematically to recognize recurring patterns, categorize challenges, and extract insights that could guide the product’s definition and design.
Key Findings
Fragmented and Inaccessible Information – Regulatory data is scattered across multiple websites, physical files, and even private email threads, leaving users without a single reliable source.
Inefficient and Frustrating Search – Existing search tools require exact legal terms and lack natural-language or topic-based search, leading to long and discouraging search processes.
Eroded Trust Due to Outdated Data – Users doubt the reliability of government sites, as many regulations are outdated or inconsistently updated, creating a sense of uncertainty.
Lack of Transparency in Published Procedures – Many official procedures are not publicly available, preventing citizens and businesses from understanding their full obligations.
Competitive & Comparative Research
As the Regulatory Database has no direct competitors, we conducted an exploratory comparative review of nine existing platforms presenting legal and regulatory information.
Each team member independently analyzed all platforms, noting subjective strengths and weaknesses, followed by a collaborative brainstorming session to identify recurring patterns and shared insights.
The goal was to learn from existing solutions—focusing on search functionality and result presentation, the two most critical components for the project’s MVP.
Key Findings
Visual Design & Layout – Clean and minimal interface using colors and icons to visually distinguish content types (e.g., ministry, status).
Accessibility & Navigation – Short tooltips for complex terms, breadcrumbs for orientation, and easy feedback options such as surveys or contact forms.
Search, Filtering & Results – Highlight document validity (active/inactive) and provide advanced filtering with sub-categories and quick clear options.
Content Page Structure – Display key document details at the top, include a side navigation menu for quick section access, and link related documents internally.
Task Analysis
Based on insights from user interviews, competitor research, and initial product requirements provided by the product manager, we evaluated and prioritized key user tasks.
Each task was rated on a 0–5 scale across three criteria, along with a summary column, to determine its relative importance and ensure that the MVP focuses on the most essential and high-impact features.
The detailed prioritization results are presented in the following table.
Findings
Card Soarting
key takeaway from the product team meeting was that the advanced search—the system’s most critical task—had unclear criteria.
Participants from legal and regulatory fields struggled to understand which items belonged to each category, so we conducted an online open card sorting study with 15 participants using UXtweak.
The results were analyzed using a Matrix of Matching (MMM) to identify optimal groupings and naming conventions.
Findings
Table kept in Hebrew to match the advanced filter labels shown in the screens
Designing the Solution
This stage translated the research insights into a practical, structured design process.
It included defining problem statements and opportunities (How Might We), shaping the system’s initial information architecture, mapping user flows, and developing both low- and high-fidelity wireframes.
Each step focused on refining the user experience and preparing the system for final prototyping.
How Might We?
At this stage, after defining the two primary user segments, we conducted an in-depth analysis of system needs through a collaborative brainstorming session in FigJam.
The process, based on insights from user interviews and competitor research, began by defining each need in the format “the user needs… in order to…”.
We then applied the How Might We methodology to reframe these needs into open-ended questions that turned problems into actionable design opportunities. This structured approach bridged the gap between understanding user needs and defining the key features for the MVP.
Findings
Professional Users – Require precision, efficiency, and flexibility to perform different types of searches. The solution includes advanced tools allowing a choice between semantic search (for contextual, broad research) and specific search (for precise document retrieval).
Business Owners and Citizens – Need clear answers to practical questions without legal knowledge. The solution provides a guided, intuitive flow that “translates” regulatory content into plain language and delivers clear, actionable results.
Information Architecture
At this stage, we created an initial information architecture focused solely on the MVP foundations.
The goal was to define the system’s core structure — the essential pages, their hierarchy, and navigation logic — ensuring clarity, scalability, and alignment with user priorities identified in earlier stages.
User Flow
Following the previous research and definition of user needs, we designed a unified user flow that accommodates both professional and non-professional users.
The flow represents the most fundamental interaction in the system — how a user searches for a document, reaches it through different search methods, and performs essential actions on it. While the system supports multiple types of searches, including advanced, generative, and categorical, all paths are unified under a single logic that ensures consistency and efficiency.
This flow served as the foundation for defining the MVP’s navigation and user interactions.
Low-Fidelity Wireframes
High-Fidelity Wireframes
Design System Pivot
In early stages, we were told the system—though governmental—could stand on its own visually and need not follow GOV design. We therefore defined an independent style guide (colors, hierarchy, typography, spacing, grid, interactive components) aligned with our wireframes and goals.
However, near the end of the high-fidelity phase and prior to building the final prototype, a meeting with the PM and the National Digital Unit clarified a policy requirement: all government systems must use the official Government Digital Design System.
We therefore pivoted from our custom guidelines to the governmental design system.
Pros
Consistency with other government platforms.
Increased user trust.
Ready-made and accessible components.
Significantly easier development process.
Cons
Required redesign of components and screens.
Reduced design flexibility.
Relearning a new system.
Results and Impact
The project was completed successfully, meeting and even exceeding the MVP goals we initially defined. Despite limited ongoing guidance from the Regulation Authority, our team managed the process independently — establishing our own workflow, structure, and weekly meetings to ensure consistent progress.
Through deep self-learning of the regulatory domain and collaborative teamwork, we transformed a complex and unfamiliar content area into an accessible, user-centered system. The mentorship and professional feedback we received from our instructor helped refine our UI and maintain a high design standard throughout the process.
I remain in communication with the product manager regarding the product’s next stages toward implementation and development.
Screens and Theoretical Foundation
The system’s design was guided by core cognitive and behavioral psychology principles, ensuring that every interaction element supports user understanding, efficiency, and trust. Below is a summary of the main theoretical concepts and their application across the key screens.
Cognitive Load Theory (Sweller, 1988) – Applied in the Home Screen Interface through clean layout, minimal visual clutter, and progressive disclosure of advanced options to prevent mental overload.
Working Memory Model (Baddeley & Hitch, 1974) – Reflected in the Advanced Search and Results Screen, where selected filters and contextual indicators reduce the need to recall previous actions, and in the Document View via breadcrumbs that support spatial memory.
Dual-Process Theory (Kahneman, 2011) – Implemented across the Search Flow, offering both intuitive, quick-access search (System 1) and detailed, parameter-based search (System 2), serving the needs of both professional and non-professional users.
Recognition Over Recall (Nielsen) – Integrated in Search Forms and Filter Menus by using dropdowns, autocomplete, and predefined terms so users can recognize relevant choices instead of recalling complex terminology.
Mental Models Alignment – Embedded in Search and Results Pages, using familiar patterns (magnifying-glass icon, search bar positioning, result-list structure) to match user expectations and minimize disorientation.
Gestalt Principles (Wertheimer, 1923) – Applied in the Results Screen, grouping information by proximity and color-coded similarity to enable fast differentiation between document types (e.g., law, regulation, procedure).
Scanning Patterns (F-Pattern; Nielsen, 2006) – Used in both Results and Document View, structuring titles and metadata along natural eye-movement paths to support efficient information scanning.
Scenario 1 — Citizens & Business Owners
Scenario 2 — Professional Users