16 KiB

Raw Permalink Blame History

Dealspace AI - Complete Lovable Prototype Specification

Comprehensive analysis from Chrome DevTools extraction and Lovable development conversation

Executive Summary

Dealspace AI is a sophisticated M&A data room platform with AI-powered diligence assistance. Built on Supabase with React/TypeScript frontend, it provides role-based access for sellers and buyers with comprehensive deal management, document organization, and Atlas AI integration for intelligent diligence support.

Demo Data Overview

The platform contains rich demo data with:

4 Deals: Project Aurora, Beacon, Catalyst, Delta
12 Documents organized in 5 folders per deal
23 Diligence Requests across 2 deals
8 Contacts with buyer/seller/advisor types
Demo Organization: "Apex Capital Partners"
Demo Users:
- Sarah Chen (Seller/Owner role)
- James Park (Buyer/Viewer role)

Architecture

Frontend (React 18 + TypeScript)

src/
├── pages/           # Main application pages (11 pages)
│   ├── Analytics.tsx     # Deal performance metrics
│   ├── AuditLog.tsx     # Activity tracking
│   ├── Auth.tsx         # Authentication with demo modes
│   ├── Contacts.tsx     # Contact management
│   ├── Dashboard.tsx    # Main overview with deal cards
│   ├── DealRoom.tsx     # Individual deal room with tabs
│   ├── DealRooms.tsx    # Deal listing with timelines
│   ├── ICMemos.tsx      # Investment committee memos
│   ├── RequestList.tsx  # Diligence request management
│   ├── Settings.tsx     # User/org settings
│   └── PlatformGuide.tsx # Documentation
├── components/      # Core reusable components
│   ├── AppLayout.tsx        # Main app shell
│   ├── AppShell.tsx         # Navigation wrapper
│   ├── AtlasGlobalChat.tsx  # AI chat interface
│   ├── AtlasSidebar.tsx     # AI assistant sidebar
│   ├── DealRequestsPanel.tsx # Request management
│   └── NavLink.tsx          # Navigation components
├── hooks/           # Custom React hooks
│   ├── useAuth.ts           # Authentication state
│   ├── useOrganization.ts   # Org context
│   ├── useTheme.ts          # Dark/light mode
│   └── useUserRole.ts       # Role-based access
├── lib/             # Utilities
│   ├── api.ts              # API layer
│   └── utils.ts            # Helper functions
├── integrations/supabase/   # Backend integration
│   ├── client.ts           # Supabase client
│   └── types.ts            # TypeScript types
└── components/ui/   # Shadcn/UI components (50+ components)

Backend (Supabase)

supabase/
├── migrations/      # Database schema (5 migration files)
│   ├── 20260214204302_*.sql  # Initial schema
│   ├── 20260214210921_*.sql  # Demo data seeding
│   ├── 20260214212128_*.sql  # Request list enhancements
│   ├── 20260214222633_*.sql  # Buyer group segmentation
│   └── 20260214224626_*.sql  # Role-based access
└── functions/       # Edge functions (5 functions)
    ├── atlas-chat/          # AI document Q&A
    ├── compute-deal-score/  # Deal scoring algorithms
    ├── demo-login/          # Demo account creation
    ├── folder-summary/      # AI folder summaries
    └── generate-ic-memo/    # IC memo generation

Database Schema (Inferred from Development History)

Core Tables

organizations - Multi-tenant organization structure
profiles - User profiles linked to Supabase auth
deals - Deal rooms with metadata
folders - Document organization within deals
documents - File storage and metadata
diligence_requests - Request list items
contacts - Contact management
audit_logs - Activity tracking

Key Fields Added During Development

diligence_requests:
- buyer_comment (text, nullable) - Buyer notes on requests
- seller_comment (text, nullable) - Seller notes on requests
- buyer_group (text, nullable) - Segmentation by buyer groups
- atlas_status (enum) - fulfilled/partial/missing
- atlas_note (text) - AI assessment summary
- confidence_score (numeric) - AI confidence rating
deals:
- ioi_date (date) - Indication of Interest date
- loi_date (date) - Letter of Intent date
- exclusivity_days (integer) - Days remaining in exclusivity
- stage (enum) - Deal stage tracking

Row Level Security (RLS)

Owner Role (Sellers): Full access to organization data
Viewer Role (Buyers): Filtered access based on buyer_group assignments
Buyer Group Filtering: Automatic data scoping based on user assignments

Atlas AI System

Core AI Functions

Atlas Chat (atlas-chat/index.ts)
- Document-based Q&A using RAG
- Concise, bullet-point responses
- Document hyperlinking (e.g., [FileName.pdf](/folder))
- Emoji-prefixed status indicators (✅/⚠️/🔴)
- Integrated with markdown parser for rich rendering
Deal Scoring (compute-deal-score/index.ts)
- Automated deal risk assessment
- Completion probability calculations
- Timeline urgency scoring
Folder Summarization (folder-summary/index.ts)
- AI-generated folder summaries
- Document completeness scoring (e.g., "85%" for Financial Statements)
- Integration with deal room folder views
IC Memo Generation (generate-ic-memo/index.ts)
- Investment Committee memo creation
- Data synthesis from deal room contents
- Buyer-role specific functionality
Demo Login (demo-login/index.ts)
- Separate seller/buyer demo account creation
- Automatic organization assignment
- RLS policy compliance for demo data access

User Roles & Access Control

Seller (Owner) Role - "Sarah Chen"

Full Platform Access:

Dashboard with all deals and metrics
Deal Rooms management (create, upload, organize)
Request Lists with "View All" consolidated view
Analytics and performance metrics
Contacts management
Audit Log visibility
Settings and configuration
Platform Guide access
Excluded: IC Memos tab (buyer-only feature)

Capabilities:

Create new deal rooms
Upload CSV request lists with optional "Buyer Group" column
Edit seller comments on requests
Manage document folders and uploads
View all buyer group request lists

Buyer (Viewer) Role - "James Park"

Limited Platform Access:

Dashboard (filtered to assigned deals only)
Deal Rooms (access only to assigned deals)
Request Lists (filtered to buyer group only)
IC Memos generation tool
Platform Guide access
Excluded: Analytics, Contacts, Audit Log, Settings

Data Filtering:

Project Aurora via "Meridian Capital" buyer group
Project Beacon via "Summit Health Equity" buyer group
No create/upload actions available
Can only edit buyer comments on requests

Specific Deal Assignments:

Project Aurora: "Meridian Capital" / "Horizon Partners" buyer groups
Project Beacon: "Summit Health Equity" / "Northstar Health Capital" buyer groups
Project Catalyst: Additional buyer groups as configured
Project Delta: Additional buyer groups as configured

Key Features Deep Dive

1. Deal Room Management

Structure:

Deal rooms with tabbed interface (Documents / Request List)
5 folders per deal with AI-generated summaries
Document upload and organization
Folder completeness scoring (e.g., "Financial Statements: 85%")

Timeline Tracking:

IOI Date (Indication of Interest)
LOI Date (Letter of Intent)
Exclusivity Period with color-coded urgency:
- Green: >30 days remaining
- Yellow: 10-30 days remaining
- Red: <10 days remaining

Deal Stages:

Stage badges replacing risk percentages
Visual status indicators
Progress tracking

2. Request List System

Segmentation:

Buyer group tabs (e.g., "Meridian Capital", "Horizon Partners")
"View All" tab for seller consolidated view
23 demo requests across Project Aurora and Beacon

Column Structure:

Request item description
Section grouping
Buyer comment (inline-editable)
Seller comment (inline-editable)
Atlas Note (AI assessment summary)
Atlas Status (fulfilled/partial/missing with confidence scores)

CSV Import/Export:

Supports "Buyer Group" column for automatic segmentation
Bulk request list management
Data interchange capabilities

Atlas Assessment:

11 Fulfilled requests with ✅ status
5 Partial requests with ⚠️ status
7 Missing requests with 🔴 status
AI-generated notes for status rationale (blank when missing)

3. Authentication & Demo System

Demo Mode Implementation:

Bifurcated demo buttons on auth page
"Seller Demo" → Sarah Chen (owner role)
"Buyer Demo" → James Park (viewer role)
Real account creation with RLS compliance
Identical data access as normal login

Security:

Supabase authentication
Organization-scoped access
Role-based navigation filtering
Comment editing permissions by role

4. Atlas AI Integration

Response Characteristics:

Concise, bullet-point format
Document hyperlinks: [FileName.pdf](/folder)
Status indicators: ✅ (complete), ⚠️ (partial), 🔴 (missing)
Markdown rendering with link support
Context-aware responses based on deal room contents

UI Integration:

AtlasGlobalChat component for document Q&A
AtlasSidebar for persistent AI assistance
Inline status indicators in request lists
Confidence scoring for assessments

5. UI/UX Features

Theme System:

Light/dark mode toggle in sidebar footer (sun/moon icon)
Defaults to dark mode
Persists via localStorage
Flash-free loading with initial script
Clean white/gray palette for light mode
Consistent teal accent color

Navigation:

Role-based sidebar navigation
Platform Guide with TOC and expandable accordions
Role badges displayed in sidebar
Responsive design for mobile/desktop

Technology Stack

Frontend:

React 18 with TypeScript
Vite build system
Tailwind CSS for styling
Shadcn/UI component library (50+ components)
Lucide Icons
State management via React Context API

Backend:

Supabase (PostgreSQL + Auth + Edge Functions)
Row Level Security (RLS) policies
Real-time subscriptions
File storage integration

AI Integration:

OpenAI API via Supabase Edge Functions
RAG (Retrieval Augmented Generation) for document Q&A
Custom prompt engineering for concise responses
Document embedding and vector search

Detailed Development History

Based on the extracted conversation, the platform evolved through these key phases:

Phase 1: Initial Demo Data Population

Rich demo data creation with 4 deals
12 documents across 5 folders per deal
23 diligence requests with AI assessments
8 contacts with type classifications
Platform Guide documentation

Phase 2: Atlas AI Refinement

Rewrote system prompts for concise responses
Added document hyperlinking capability
Implemented emoji status indicators
Enhanced markdown parser for link rendering

Phase 3: UI/UX Enhancements

Light/dark mode toggle implementation
Request list integration within deal rooms
Tabbed interface (Documents / Request List)
Responsive design improvements

Phase 4: Demo Authentication System

Demo mode implementation for testing
Real demo account creation via edge function
RLS compliance for demo data access
Bypass authentication for UI testing

Phase 5: Request List Enhancements

Buyer/seller comment columns (inline-editable)
Buyer group segmentation with tabs
CSV import with "Buyer Group" column support
Atlas Note column for AI assessments

Phase 6: Buyer Group Segmentation

Multiple request lists per deal by buyer group
Demo data for buyer group assignments
"View All" seller perspective
Filtered buyer access by group

Phase 7: Atlas Assessment Integration

Atlas Note column implementation
Brief AI summaries for request status
Blank notes for missing items
Status-based conditional display

Phase 8: Deal Timeline Features

Removed close percentage and risk metrics
IOI/LOI date tracking implementation
Exclusivity period countdown with color coding
Stage-based deal progression

Phase 9: Role-Based Access Control

Bifurcated demo login (Seller/Buyer)
Role-based sidebar navigation
Data filtering by buyer group assignments
Comment editing permissions by role
Role badge display in sidebar

Migration to Go - Implementation Guide

1. Database Migration

-- Example table structure based on features
CREATE TABLE organizations (
    id UUID PRIMARY KEY,
    name TEXT NOT NULL,
    created_at TIMESTAMPTZ DEFAULT NOW()
);

CREATE TABLE deals (
    id UUID PRIMARY KEY,
    organization_id UUID REFERENCES organizations(id),
    name TEXT NOT NULL,
    stage TEXT,
    ioi_date DATE,
    loi_date DATE,
    exclusivity_days INTEGER,
    created_at TIMESTAMPTZ DEFAULT NOW()
);

CREATE TABLE diligence_requests (
    id UUID PRIMARY KEY,
    deal_id UUID REFERENCES deals(id),
    description TEXT NOT NULL,
    section TEXT,
    buyer_group TEXT,
    buyer_comment TEXT,
    seller_comment TEXT,
    atlas_status TEXT CHECK (atlas_status IN ('fulfilled', 'partial', 'missing')),
    atlas_note TEXT,
    confidence_score DECIMAL(3,2),
    created_at TIMESTAMPTZ DEFAULT NOW()
);

2. Authentication System

JWT-based authentication with organization scoping
Role-based access control (owner/viewer)
Demo account generation endpoints
Session management with refresh tokens

3. AI Integration Architecture

// Example Go structure for Atlas AI
type AtlasService struct {
    openaiClient *openai.Client
    vectorStore  VectorStore
}

type DocumentQuery struct {
    Question string `json:"question"`
    DealID   string `json:"deal_id"`
    Context  []string `json:"context"`
}

type AtlasResponse struct {
    Answer     string  `json:"answer"`
    Confidence float64 `json:"confidence"`
    Sources    []DocumentReference `json:"sources"`
    Status     string  `json:"status"` // fulfilled, partial, missing
}

4. API Endpoints

POST /api/auth/demo-login
POST /api/atlas/chat
POST /api/atlas/folder-summary
POST /api/atlas/ic-memo
GET  /api/deals
GET  /api/deals/{id}/requests
POST /api/deals/{id}/requests/csv-import
PATCH /api/requests/{id}/comments

5. Role-Based Filtering

func FilterRequestsByBuyerGroup(requests []Request, userRole string, buyerGroup string) []Request {
    if userRole == "owner" {
        return requests // Sellers see all
    }
    
    var filtered []Request
    for _, req := range requests {
        if req.BuyerGroup == buyerGroup || req.BuyerGroup == "" {
            filtered = append(filtered, req)
        }
    }
    return filtered
}

Business Logic Patterns

Demo Data Seeding

4 predefined deals with realistic names
Buyer group assignments per deal
AI-generated folder summaries
Request list segmentation by buyer groups

Atlas AI Prompt Engineering

System prompts optimized for concise responses
Document context injection for RAG
Status classification logic (fulfilled/partial/missing)
Confidence scoring algorithms

Audit Logging

Activity tracking across all user actions
Role-based log filtering
Timeline view of deal progression
User interaction monitoring

Critical Implementation Notes

RLS Compliance: All data access must respect buyer group assignments
AI Response Format: Maintain emoji status indicators and document linking
Demo Mode: Ensure demo accounts have identical data access to real accounts
CSV Import: Support buyer group column for request segmentation
Timeline Urgency: Implement color-coded exclusivity period warnings
Comment System: Separate buyer/seller comment fields with role-based editing
Theme Persistence: Implement localStorage-based theme switching
Responsive Design: Mobile-first approach with sidebar collapsing

This comprehensive specification captures the complete feature set, data model, and implementation patterns from the Dealspace AI Lovable prototype, providing a complete blueprint for Go migration.

16 KiB Raw Permalink Blame History