Enterprise CRM Operations
Live Application: https://crm.enriquesamaniego.com · Also known as: AI-Powered CRM
Centralizes customer accounts, leads, commercial opportunities, and follow-up activities in a single cloud database protected by Row-Level Security (RLS) and automated webhook notifications.
Before you launch: the demo opens at a sign-in screen. Select “Create Temporary 24h Demo” to explore with synthetic data. No real client data is used.
Launch Live CRM AppLive web application hosted at crm.enriquesamaniego.com. Built with synthetic/demo-safe operational boundaries to protect database integrity.
Commercial teams frequently operate with disconnected sales pipelines and client records scattered across spreadsheets, email threads, and manual notes. This fragmentation causes missed follow-ups, inconsistent pipeline reporting, and weak access control over sensitive customer data.
B2B sales teams, account managers, and operations leaders who need structured pipeline tracking without enterprise licensing bloat.
Defined the commercial data model, operational stages, and access security rules; architected the Supabase PostgreSQL schema with Row-Level Security (RLS) policies; configured outbound webhook triggers; and built the web application using AI-assisted engineering in Google Antigravity.
Replaced fragmented lead spreadsheets with a centralized, access-controlled commercial database. Provides a single source of truth for opportunity status, follow-up accountability, and structured pipeline visibility.
How the Solution Works in Practice
The application consolidates contacts, accounts, deal stages, and scheduled follow-ups into a unified workspace. Every database query is scoped at the PostgreSQL engine level via Row-Level Security (RLS) policies tied to authenticated user sessions, while stage transitions trigger webhook events to keep operational teams aligned.
Authenticated Session & Role Resolution
Users sign in via Supabase Auth; session tokens attach identity claims enforced directly by PostgreSQL Row-Level Security policies.
Lead Capture & Account Structuring
Prospects, organizations, and contact histories are recorded against normalized relational tables rather than ad-hoc spreadsheet rows.
Pipeline Stage Progression
Opportunities move through defined commercial stages with structured follow-up tasks and pipeline value aggregation.
Webhook Event Dispatch
Key state changes emit structured webhook payloads to downstream automation workflows for team notifications.
Scope of My Direct Responsibility
- Product definition and commercial pipeline modeling
- Relational database schema design (PostgreSQL / Supabase)
- Authentication and Row-Level Security (RLS) policy enforcement
- Automated webhook notification integration for pipeline events
- Cloud deployment and operational validation
Architecture & Component Responsibilities
Modern Web Frontend
Responsive pipeline views, contact records, opportunity cards, and activity tracking.
Session Authentication & RLS
Enforces user authentication and row-level data isolation directly inside the database engine.
Relational Commercial Store
Normalized tables for accounts, leads, opportunities, stages, and follow-up logs.
Event Webhooks
Dispatches outbound JSON notifications when lead or deal statuses change.
Built and iterated using Google Antigravity for AI-assisted code generation, schema validation, and security review, compressing the engineering cycle from months to weeks.
Runtime operations are deterministic: authentication, PostgreSQL relational integrity, Row-Level Security (RLS), and webhook workflows. It does not rely on black-box LLM scoring to make commercial decisions.
Key Decisions, Discarded Alternatives & Trade-Offs
Decision Implemented
Enforce Row-Level Security (RLS) directly in PostgreSQL
Discarded Alternative
Application-only filtering in frontend or middleware queries
Executive & Operational Rationale
In enterprise environments, UI-level filtering leaves data exposed if an endpoint is misconfigured. Database-enforced RLS guarantees that unauthorized rows cannot be read or mutated regardless of the client request.
Decision Implemented
Use Google Antigravity for AI-assisted build on top of a standard relational schema
Discarded Alternative
No-code spreadsheet wrappers (Airtable/Glide) or multi-month manual boilerplate
Executive & Operational Rationale
Spreadsheet wrappers break down under strict permission requirements, while AI-assisted engineering on PostgreSQL and React delivers a maintainable codebase with full security governance in weeks.
Decision Implemented
Decouple alerts via outbound webhooks
Discarded Alternative
Hardcoding SMTP/email logic inside UI components
Executive & Operational Rationale
Separating core CRUD operations from notification delivery keeps the CRM responsive and allows workflows to evolve independently.
What Is Verified vs. What Remains to Be Measured
Deployed as an operational demonstration and executive engineering case. Qualitative improvements in data structure and access control are verified; commercial conversion percentages depend on organization-specific sales volume and are not claimed here.
Current Scope & Guardrails
- Designed as a focused B2B pipeline and account tracker rather than a full ERP/billing suite.
- Public live application access is scoped to prevent destructive operations or exposure of private customer records.
Planned Next Improvements
- Add automated lead-enrichment webhook workflows using verified company domain lookups.
- Introduce read-only demo role presets with scheduled 24-hour synthetic data reset.
Related Workflows & Operational Learnings
Hermes Agent on VPS: Multi-Channel Operational Workflows
Task-specialized profiles with explicit input/output boundaries outperform generic open-ended prompts. Running the agent platform on a dedicated VPS provides full control over logs, environment secrets, and channel access lists.
Read Lab Note →Explore the Other Deployed Case Studies
Project Management Platform (ProjectFlow)
Cross-functional projects, tasks and milestones with role-based access and automated blocker alerts.
Before you launch: the demo opens at a sign-in screen. Select “Create Temporary 24h Demo” to explore with synthetic data. No real client data is used.
Intelligent Web Engine (TimberFlow)
B2B website with adaptive messaging, edge caching and a rate-limited OpenAI integration.
Before you launch: public sample site, no sign-in required. Content is for demonstration only.
Interested in the engineering or governance behind Enterprise CRM?
Let's discuss how I apply this operational discipline to enterprise IT, Cloud, and AI leadership roles.