Operational Dashboard
KPI cards, a focus banner for priority issues, active transactions table, priority tasks panel, and a recent activity feed. Admins see across all agents; agents see their own book.
A transaction management platform for real estate agents and brokerages. Tracks every deal from contract to close: deadlines, documents, compliance, and tasks, without the chaos of spreadsheets and sticky notes.
Built with Jenny Florey
Stack: Next.js, Supabase, TypeScript, Tailwind
Status: Beta with active brokerage client
Context
Real estate agents are managing a stack of moving parts on every transaction: contingency deadlines, required documents, compliance checklists, buyer and seller communication, title company coordination, and the daily reality that one missed date can blow a deal.
The tools that exist are either too generic or too expensive for smaller brokerages to run on every agent. ClosePilot was built to fill that gap, specifically for the Gulf Coast market, with Alabama-specific deadline logic built in rather than bolted on as an afterthought.
Jenny brought domain authority here. She is an active realtor who knows exactly what agents actually need to see at 8am when they are juggling four active transactions. That context shaped every screen.
Functionality
ClosePilot is organized around transactions: one record per deal, from the moment a contract is signed through closing. Everything else hangs off that central record.
KPI cards, a focus banner for priority issues, active transactions table, priority tasks panel, and a recent activity feed. Admins see across all agents; agents see their own book.
Full detail view per transaction: executive snapshot, top issues, party management, milestones, task list, documents, compliance checklist, activity feed, and timeline.
Deadlines are calculated from contract data using Alabama business day rules, with Alabama Code § 1-3-8 holidays implemented directly.
Upload a contract PDF and the system extracts structured data: dates, parties, key terms. Document type is auto-classified using keyword rules and transaction-type-aware defaults.
Each transaction type has a defined document requirement set. The system tracks what has been uploaded, flags what is missing, and updates compliance status automatically.
Global task view and per-transaction task lists. Tasks carry priority, status, and due dates, link to transactions, and roll up into the dashboard focus panel.
Cmd/Ctrl+K opens a search modal across transactions, parties, tasks, and documents simultaneously.
Brokerage branding, SMTP configuration, user management, demo mode, audit log viewing, and a customization inquiry system.
Layout & UX
The people using ClosePilot every day are agents trying to close deals. The design principle throughout was: surface the right thing at the right time, do not make them hunt.
At the top of the dashboard, a focus banner aggregates the things that actually need attention right now: missing required documents, high-risk flags, overdue tasks.
| Role | What they see |
|---|---|
| Agent | Their own active transactions, tasks, and documents. Full transaction workspace for each deal. |
| Admin | Cross-agent visibility, user management, brokerage configuration, audit log, system settings, and demo controls. |
| Transaction Coordinator | Compliance checklists, document status, and deadline visibility across the book. |
First-time users get a guided in-app tour, plus a dedicated Getting Started page. The Help page includes Alabama-specific workflow guidance.
Each transaction has a full-featured detail page: not a form, a workspace. A printable view generates a clean summary for sharing with clients or keeping on file at close.
Security & Data
Real estate transaction data is sensitive. Client financials, personal information, property details, contract terms: none of it should be visible to anyone other than the agent and brokerage who own it. ClosePilot enforces that at the database layer, not just in the application layer.
Every table in the database has Row Level Security enabled via Supabase. Users can only read and write their own data. This is enforced by Postgres policy, not by application logic. Admins get broader access through explicit policy grants, scoped to their brokerage.
Supabase Auth handles email/password login, password reset, auth callbacks, and session management. Middleware enforces route protection. Disabled accounts are enforced through a profiles.is_active flag.
Personal API keys stored on user profiles are encrypted at rest using a configurable 32-byte encryption key. The service role key never touches the browser.
Auth endpoints and API routes are rate-limited at the middleware layer. Security headers are set globally.
Every significant action is written to an audit log: who did what, when, on which record.
ClosePilot document processing is implemented using local PDF parsing, keyword rules, and regex extraction. The shipping document automation in the current version is deterministic, not LLM-driven. That is deliberate for beta: predictable, auditable, no external API dependency for core workflow logic.
Technical Architecture
Next.js 14 App Router with TypeScript throughout. Supabase handles authentication, the PostgreSQL database, and file storage. The schema evolves through versioned migrations. UI components are built on Radix UI primitives via shadcn, styled with Tailwind CSS.
The server action pattern is used consistently: all database writes go through server actions, never directly from client components. Alabama-specific business day logic implements Alabama Code § 1-3-8 holidays directly.
Test coverage is focused on the domain logic that matters most: deadline engine, timeline logic, risk flag rules, and date utilities. These are the calculations that agents depend on.
Notes
ClosePilot is a case study in building operational software for a domain-specific audience, not a generic project management tool with a real estate skin. The Alabama business day implementation, the transaction-type-aware document requirements, and the Gulf Coast MLS workflow guidance come from building with someone who does this work daily.
The beta deployment with a local brokerage in Fairhope is the real test. Software that agents actually use looks different from software that demos well.
If you are building workflow software for a specialized professional audience, real estate, legal, construction, healthcare, I'd be glad to talk about how domain expertise changes what gets built.
If you are building something that needs domain judgment, careful systems thinking, or practical AI support, send me a note.