[ 04 // PROJECT SCOPE & RESULTS ] · CLIENT: Confidential GovTech Venture [NDA] (Emerging Markets & Global) · ROLE: Product engineering and system architecture · 2026 · MVP validated; next phase in development

GovTech: a platform for infrastructure documents and obligations

From validating a document workflow to designing a dedicated system for files, access control and AI-assisted obligation extraction.

VALIDATION Initial MVP validated

CURRENT STAGE Dedicated platform in development

DATA ACCESS PostgreSQL + row-level rules

AI SCOPE Obligation extraction for review

[ 01. BUSINESS CONTEXT ]

What needed to change

Public-private infrastructure projects involve several organisations working with agreements, technical files and long-term obligations. The venture needed to validate how those users would organise documents and manage access before committing to a dedicated platform. The first MVP tested the core workflow. The next stage addresses larger files, more explicit data relationships and access rules. The client’s identity and confidential contract details remain undisclosed.

[ 02. SYSTEM ARCHITECTURE ]

How the system fits together

[ Phase 1: workflow validation ]
    Webflow + Memberstack + Airtable + Make + Cloudflare R2
                         │
[ Phase 2: architecture under development ]
    Next.js application
        ├── Authorised file upload → Cloudflare R2
        ├── Authentication and data → Supabase / PostgreSQL
        └── Document processing → Python / FastAPI / LLM
                                      │
                              Extracted obligations → human review

[ 03. ARCHITECTURAL DECISIONS ]

Scope and key decisions

1. Validate the workflow first. The initial system combined Webflow, Memberstack, Airtable and Make to test the user journey. This provided a basis for the next architecture decision. 2. Separate files from form limits. We designed a Cloudflare Workers and R2 upload path for large documents, followed by direct uploads using time-limited signed URLs in the next architecture. 3. Define data access explicitly. The dedicated platform design uses Supabase and PostgreSQL with row-level access rules so permissions can be expressed alongside the data model. 4. Keep AI analysis a separate service. Python and FastAPI support a document-processing pipeline for extracting obligations and milestones. Extracted information is intended for human review. This is the scope of the developing system, not a claim of completed rollout.

[ 04. RESULTS & DELIVERY STATUS ]

What the work delivered

▶ The initial MVP validated the document workflow and informed the next development phase.

▶ The architecture separates document storage, application data, access control and AI analysis.

▶ A dedicated document-processing service is being developed to support obligation extraction.

When this approach applies

Relevant when an organisation has validated a process and needs to turn it into a dedicated platform with document storage, multiple user roles and controlled AI assistance.

[ 05. TECHNOLOGIES & TOOLS ]

Next.js 14 TypeScript Supabase (PostgreSQL) Cloudflare Workers Cloudflare R2 Python FastAPI Tailwind CSS Figma Design System

[ A SIMILAR CHALLENGE? ]

Discuss your system

Describe the process, current constraints and result you need.

Discuss your project