X11 Case Study: Digital Platform for Investor Management
Brief description
AI lawyer web service: an online legal consultant across jurisdictions powered by LLM, with its own knowledge base, subscriptions, and chat history.Task
Build a product that gives users fast legal answers across jurisdictions through a specialized AI assistant, with authorization, pricing plans, and saved inquiry history.About the project
We built a legal consultation service on top of AI model APIs. For the task we wrote a custom neuro-lawyer prompt and assembled a knowledge base for legal questions across jurisdictions. We delivered a full UI system with authorization, subscription plans, and stored user chat history.Business benefits
•MARKET LAUNCH WITH A READY MVP IN 3 MONTHS BY INTEGRATING AI MODEL APIS INSTEAD OF TRAINING A PROPRIETARY MODEL FROM SCRATCH
•UP TO 80% BUDGET SAVINGS AT THE START — A CUSTOM PROMPT AND OWN KNOWLEDGE BASE DELIVER SPECIALIZED AI-LAWYER QUALITY WITHOUT THE COST OF TRAINING A MODEL
•MONETIZATION FROM DAY ONE — PRICING PLANS AND SUBSCRIPTIONS ARE BUILT INTO THE PRODUCT, PAID ACCESS TURNS ON WITHOUT EXTRA WORK
•SCALE WITHOUT REBUILDING THE PRODUCT — THE KNOWLEDGE-BASE ARCHITECTURE SUPPORTS NEW JURISDICTIONS AND A FUTURE MOVE TO A PROPRIETARY AI MODEL
Features
Registration and personal account
•Fast user registration via email or phone
•Authorization and access to the personal account
•Store and delete user inquiry history
•Access to saved dialogues
•Manage basic profile data
Legal chat interface
•Messenger-style interface for talking to the AI lawyer
•Send text legal questions
•Attach files to an inquiry
•Save full dialogue history in the personal account
•Separate new and past consultations
•Continue a previous dialogue without losing context
Legal request processing
•Understand the context of the user’s inquiry
•Identify the legal problem from the message
•Automatic clarifying questions
•Collect missing information before forming an answer
•Recognize legally significant entities: names, dates, organizations, amounts, statute articles, case circumstances
•Form an answer using the user’s input
AI core and answer logic
•Configure AI rules through prompt engineering
•Create basic legal consultation scenarios
•Configure answer limits and structure
•Build clarifying-question logic
•Support a dialogue consultation scenario
•Architecture prepared for a later move to a proprietary AI model
Legislative knowledge search
•Build an internal legal document base
•Search legislative materials inside the platform
•Use the internal base as the priority data source
•Work with codes, laws, regulations, and other normative materials
•Structure prepared to expand with case law and legal doctrine
Working with legal materials
•Upload user files into the chat
•Pass files into processing within the consultation scenario
•Use attached materials as extra inquiry context
•Store files within a specific dialogue
Subscription system
•Paid plans can be connected
•Access-limiting logic by subscription prepared
•Further user segmentation by pricing plans supported
•Limits on inquiry count, features, or processing volume
Payment integration
•Ability to connect a payment provider
•Support for YooKassa, CloudPayments, Stripe, or similar solutions
•Payment logic for subscriptions or individual consultations prepared
•Payment status handling inside the platform
Technology stack
1. Frontend:
•React Native + Expo — single codebase for iOS, Android, and Web
•React 19 + TypeScript — modern type-safe frontend
•Expo Router — file-based routing and navigation
•TanStack React Query — caching, sync, and data updates
•NativeWind (Tailwind CSS) — utility-first styling
•React Hook Form — auth, registration, and settings forms
•Axios + Orval — typed API client and OpenAPI codegen
2. Backend:
•ASP.NET Core 8.0 — server application foundation
•ASP.NET Core Identity + JWT — authentication
•Entity Framework Core 9.0.9 — ORM
•Hangfire — background jobs
•MailKit and MimeKit — email
3. API:
•MediatR — CQRS for queries and commands
•Swashbuckle — OpenAPI/Swagger docs
•Asp.Versioning — API versioning
4. Integrations:
•AWS SDK — Amazon S3
•Stripe.net — payments
•OpenAI — artificial intelligence
5. Databases:
•PostgreSQL — primary database
6. Development tools:
•Docker — containerization
•Serilog — logging and monitoring
Working with the client
*Important: this is not project delivery work — it is work with the client specifically, and it is not in the estimate or commercial proposal; it is what we do to raise the product’s chance of successKnowledge that makes the client independent and resilientQuality code is the baseline. What really changes a project’s fate is the client’s ability to make the right decisions without us. We handed the client’s team not just a platform manual, but an understanding of its business logic: how the product economy works, where failure points are, how to read analytics and react in time. We ran training, modeled several crisis scenarios, and showed with live examples how to stay resilient when the market does not follow the plan. As a result the client cut launch costs, avoided typical beginner mistakes, and entered the market with a systemic view rather than a set of guesses.Our goal is not just development — it is our client’s success.Design goals:
•REMOVE THE BARRIER “LEGAL HELP IS HARD AND EXPENSIVE”: A CONSULTATION SHOULD FEEL LIKE CHATTING IN A FAMILIAR MESSENGER
•TAKE THE USER FROM SIGN-IN TO THE FIRST QUESTION IN AS FEW STEPS AS POSSIBLE TO RAISE FIRST-DIALOGUE CONVERSION BY 20%
•SHOW PLAN VALUE TRANSPARENTLY SO THE MOVE TO A SUBSCRIPTION IS INTENTIONAL
Design results:
•FAMILIAR CHAT UI: THE USER ASKS THE FIRST QUESTION RIGHT AFTER SIGN-IN, WITH NO PRODUCT TRAINING
•THE AI LAWYER’S CLARIFYING QUESTIONS FEEL LIKE A NATURAL DIALOGUE — A LIVE CONSULTATION SENSATION GROWS, AND TRUST IN ANSWERS RISES
•INQUIRY HISTORY IS ALWAYS AT HAND: PAST CONSULTATIONS ARE EASY TO REOPEN, WHICH DRIVES REPEAT USE
•PLANS AND LIMITS ARE SHOWN CLEARLY: THE USER SEES WHAT THE SUBSCRIPTION GIVES BEFORE PAYING
