Back to work

Business software · Applied AI

ZIGOMA

AI-native ERP for modern SMEs

An integrated business system connecting CRM, sales, invoicing, stock, projects and finance, with contextual AI governed by permissions and human validation.

Period2026
My roleLead Developer · Full-stack engineering · Software architecture · Applied AI · Product engineering · UX decisions · Testing strategy

Python · Flask · Firestore · OpenAI · Pandas · NumPy · Tailwind · WeasyPrint

ZIGOMA composition with ERP dashboard, CRM pipeline, invoicing and projects.
01

Context

ZIGOMA is a modular business ERP designed to bring an SME’s daily operations into one product. It connects customer relationships, sales, invoicing, stock, projects and finance.

02

Business problem

When those activities live in separate tools, data is entered repeatedly, workflows become manual and operational visibility suffers. The product aims for continuity across CRM, quotes, orders, delivery, invoicing and financial follow-up.

03

My role

  • Lead Developer
  • Full-stack engineering
  • Software architecture
  • Applied AI
  • Product engineering
  • UX decisions
  • Testing strategy
04

Product & solution

The product organizes business modules around connected workflows: CRM, sales, quotes, orders, delivery notes, invoicing, stock, projects, finance, analytics and an AI assistant. Sensitive actions remain subject to roles, permissions and human validation.

05

Inside the product

ZIGOMA — Indicators and activity across ERP domains.

01 / 06Dashboard

Indicators and activity across ERP domains.

06

Architecture

Layered boundaries allow business domains to evolve without mixing interface, rules, storage and integrations. This is a conceptual view and exposes no private code.

  1. User interface and experience
  2. Application and orchestration
  3. Business services and modules
  4. Data and consistency controls
  5. Contextual AI and integrations
07

AI layer

AI works within business context: it can help interpret, structure and propose, but cannot bypass permissions or validation. BASE and voice provide additional ways to interact.

  • Business context and targeted capabilities
  • BASE and voice interaction
  • Structured reasoning
  • Permission checks
  • Human validation
  • Guardrails and traceability
08

Engineering challenges

Double-entry accounting

Preserving balanced entries and consistent financial flows.

Idempotence

Preventing repeated actions from producing duplicate business effects.

Accounting period controls

Respecting accounting periods in financial operations.

Atomic stock transfers

Keeping both sides of a transfer consistent.

Multi-company isolation

Separating data and actions by the active company.

RBAC

Aligning available actions with roles and permissions.

Anti-double billing

Protecting document flows against duplicate invoices.

Auditability and testing

Making business decisions and transitions verifiable.

09

Key capabilities

  • CRM and sales tracking
  • Quotes, orders and delivery notes
  • Invoicing
  • Stock movements
  • Project management
  • Finance and accounting
  • Dashboards
  • Contextual business assistant
  • Multi-company and RBAC
10

Technology stack

  • Python
  • Flask
  • Firestore
  • OpenAI
  • Pandas
  • NumPy
  • Tailwind
  • WeasyPrint
11

Outcomes

  • A connected business flow from customer relationship to invoice.
  • Financial and stock operations designed to remain controllable.
  • AI assistance integrated into the product with human validation.