Bops
A team of AI bots that runs your business ops: inbox, pipeline, invoices, reports. Each one has its own computer, email and phone number, and remembers everything.
翻訳待ち
Facil — Scalable AI-Powered Government Digital Services Framework. Document intelligence, workflow automation, compliance monitoring. Currently deployed for Equatorial Guinea.
翻訳待ち
Loading the full README. The saved excerpt is shown below.
Powering digital transformation for government procedures, workflows, and fiscal services — deployable to any country.
Currently deployed for the Republic of Equatorial Guinea.
Facil is a full-stack, AI-powered framework for digitizing government administrative procedures. It provides a complete platform — backend API, web dashboard, citizen mobile app, and field inspection app — that any government can configure to manage its own fiscal services, workflows, document processing, and citizen interactions.Read the full README ↗Powering digital transformation for government procedures, workflows, and fiscal services — deployable to any country.
Currently deployed for the Republic of Equatorial Guinea.
Facil is a full-stack, AI-powered framework for digitizing government administrative procedures. It provides a complete platform — backend API, web dashboard, citizen mobile app, and field inspection app — that any government can configure to manage its own fiscal services, workflows, document processing, and citizen interactions.
The framework is country-agnostic by design: workflows, fiscal services, entities, roles, translations, and business rules are all data-driven and configurable through the database and admin UI — not hardcoded. Deploying Facil to a new country means configuring its services, entities, and workflows, not rewriting code.
Current deployment: Republic of Equatorial Guinea — 873 fiscal services, 21 ministries, 20 government entities, 36 business workflows, trilingual (Spanish, French, English).
End-to-end digitization of administrative procedures. 36 predefined workflows (passport, residency, driving license, contracts, inspections...) with configurable steps, dynamic form validation, conditional routing, and multi-entity approval chains. Citizens submit requests through guided wizards; government agents process them via role-based dashboards with SLA-driven priority queues, automatic assignment, and real-time workload balancing across 20 government entities.
A RAG chatbot powered by Gemini 2.5 Flash with role-based specialized agents — each agent has domain-specific tools and access controls respecting strict data confidentiality:
| Agent | Scope | Tools | Confidentiality |
|---|---|---|---|
| Citizen Agent | Public services, fees, procedures | 19 public tools | No access to internal data |
| Authenticated Agent | Personal requests, status tracking | 8 auth tools | Own data only |
| Treasury Agent | Payment validation, revenue analysis | Deep reasoning | Entity-scoped financial data |
| Supervisor Agent | Team performance, anomaly detection | Deep reasoning | Entity-scoped agent metrics |
| Admin Agent | System configuration, audit analysis | Deep reasoning | Full system access with audit trail |
Each agent operates with Chain-of-Thought reasoning (6-step prompt), self-reflection (auto-regeneration if quality score < 5/10), and 15 response formats.
Convert physical documents (contracts, invoices, identity papers, government forms, tax declarations) into structured digital data. The OCR pipeline extracts, validates, and structures data from 39+ document schemas using template-based field mapping with bounding box annotation, AI-powered classification, and cross-document validation. Extract structured text from PDFs while preserving layouts.
Every agent action, payment validation, document review, and permission change is logged in the audit trail (2,800+ entries). The SchemaValidationEngine enforces document integrity with 70+ JSON rules. RiskAnalyzer runs a 12-step assessment pipeline on every document. All API requests pass through rate limiting, JWT validation, RBAC check (335 permissions), and parameterized SQL — ensuring OWASP compliance at every layer.
AI-driven analysis: detect recurring patterns across scanned documents, validate data by cross-referencing multiple sources (OCR extraction vs form input vs database records), and enable enterprise-grade search across the fiscal services catalog (873 services, 21 ministries). Automatic zone resolution with Levenshtein fuzzy matching and commerce type classification for commercial license workflows.
| Metric | Value |
|---|---|
| Fiscal Services Catalog | 873 (869 active) |
| Government Entities | 20 |
| Business Workflows | 36 predefined |
| Database Tables | 145 |
| Database Enums | 50 |
| Roles | 47 (335 granular permissions) |
| OCR Document Schemas | 39 |
| Languages | 3 (ES, FR, EN — extensible) |
| Backend Modules | 30 |
| Frontend Modules | 41+ |
| Communication Templates | 85 (29 email, 31 SMS, 9 push, 16 in-app) |
| Audit Log Entries | 2,800+ |
| Total Commits | 3,800+ |
This is the beginning of the README. Read the rest at the original. Read the full README ↗