Odoo vs ERPNext vs AI-built ERP — every angle from a 20-year dev

I've been writing software for over 20 years and have seen many ERP generations — from SAP and Oracle down to small open-source tools. The question I keep getting in the last 2–3 years is "Odoo or ERPNext — which should we pick?" followed by a newer one: "What about having AI write our own ERP?" This article compares everything — features, UI/UX, development, maintenance, cost, upgrade policy, and AI-built ERPs — all in one read. For each angle I'll tell you what's real and what's just marketing.

Note: based on documentation and real-world experience as of November 2026 — Odoo 20 and ERPNext v16. Details may change with versions and configurations. This is a technical opinion piece, not legal or accounting advice.

1. The big picture — different DNA from day one

Odoo is the world's largest open-source ERP. It started as a CRM and grew into a suite of 80+ modules. It has two editions: Community (free, LGPL) and Enterprise (paid, with extra modules). Its strength is having everything in one place — CRM, accounting, manufacturing, e-commerce, POS, HR, website builder — like "WordPress for ERP." You install it and get everything, then enable only what you need.

ERPNext is built on the Frappe Framework (Python). It's fully open source under GPL v3 — no paid edition. Its strength is being a "real ERP" from the start — accounting, inventory, manufacturing, HR, CRM, projects were all built specifically for serious business use, not bolted on later.

The most important difference:Odoo is "a platform that happens to contain an ERP"; ERPNext is "an ERP that happens to be a platform." The result: Odoo is strong on breadth (many features), while ERPNext is strong on depth (deeper ERP in each module).

2. Feature comparison, module by module

2.1 Accounting

Odoo: accounting exists in both editions — Community has basic journals, Enterprise has full accounting built for real compliance. Since v18 the accounting was completely reworked — bank reconciliation, deferred revenue, asset management. Watch out for local tax (VAT, withholding) per country — Odoo supports many but needs correct configuration.

ERPNext: its original strength — chart of accounts, journal entries, payment reconciliation, financial statements (trial balance, P&L, balance sheet) are deep and detailed. Withholding tax, multi-currency, deferred accounting supported. Thai tax reports have some community help but need customization.

Winner: ERPNext — accounting is its core, deeper and fully open source. Odoo's accounting improved a lot in v18+ but full features need Enterprise, and Thai local tax still needs customization.

2.2 Inventory

Odoo: very strong — multi-warehouse, batch/serial tracking, putaway rules, barcode, multiple valuation methods (FIFO/average/standard). Suits both retail and manufacturing.

ERPNext: equally strong — immutable stock ledger entries (can't edit, only reverse), serial/batch tracking, quality inspection, stock reconciliation. The append-only ledger gives a crystal-clear audit trail, great for compliance.

Tie: both excellent — Odoo favors flexibility, ERPNext favors auditability.

2.3 Manufacturing

Odoo: complete MRP — BOM, work orders, work centers, unbuild orders, subcontracting, shop floor control. The planning engine improved significantly in v18+. Suits factories of all sizes.

ERPNext: has BOM, work order, job card, production plan, but less deep — especially advanced planning and the shop-floor UI that Odoo built specifically for floor workers.

Winner: Odoo — manufacturing is a real strength, and its shop floor app is much better for floor staff.

2.4 Sales / CRM / POS / E-commerce

Odoo: clearly ahead — CRM pipeline, quotation, sales order, subscription, rental, POS (offline-capable), e-commerce website builder, marketplace integrations. Ideal for multi-channel selling.

ERPNext: has sales order, quotation, customer portal, basic POS — but no built-in e-commerce; integrate separately. Better for B2B than heavy B2C.

Clear winner: Odoo — if you sell across channels, built-in POS + e-commerce wins by a mile.

2.5 HR / Payroll

Odoo: good HR app — employee records, time off, attendance, recruitment, appraisals. But payroll isn't in Community — use a third-party module or Enterprise (paid). Thai local payroll needs heavy customization.

ERPNext: complete HR — employee, attendance, leave, payroll engine (real payroll processing). Supports salary structures, tax slabs. Thai payroll (social security, tax, leave) still needs customization. Neither is 100% "ready" for Thailand.

Tie: both need Thai customization; ERPNext's payroll engine is slightly deeper.

2.6 Project / Helpdesk / everything else

Odoo has everything — project, helpdesk, website, forum, marketing automation, social media, email marketing — breadth ERPNext doesn't match. ERPNext has project and support tickets, but not as wide.

Winner: Odoo for "complete suite."

3. UI/UX comparison

Odoo: very modern UI since v17+ — OWL components + SCSS, beautiful, responsive, feels like a modern SaaS. Real iOS/Android apps that actually work well. The UX is well thought out — form, list, kanban, calendar views switch easily, and end users don't need much training.

ERPNext: usable UI but more "functional" — Frappe Desk puts lots of data on one screen. Good for people who like seeing everything at once, but can feel dense for newcomers. Mobile experience isn't as polished as Odoo's.

Winner: Odoo — noticeably better UI/UX, especially if end users aren't back-office staff.

4. Development and customization

Odoo: Python + XML/OWL (frontend). Its ORM is highly abstract — _inherit a model and override fields or methods directly. Great for Python devs. Downside: upgrades sometimes require fixing custom code, and OWL frontend has a steeper curve than Vue/React.

ERPNext: Python + Frappe Framework. DocType system — define a document type and get CRUD + API + permissions automatically (similar concept to Ash but in Python). Client scripts (JS) for UI logic, server scripts (Python) for business logic. Great for devs who like clearly enforced structure. Downside: heavy framework lock-in — hard to customize if you don't understand Frappe.

Example of a custom field in ERPNext (DocType JSON):

{
  "doctype": "Custom Field",
  "dt": "Sales Order",
  "fieldname": "loyalty_points",
  "fieldtype": "Int",
  "label": "Loyalty Points"
}

vs Odoo where you write a Python model + XML view + security rules separately — ERPNext is easier here. But for complex logic, Odoo is more flexible.

Winner: depends on the team — Python devs who like clear structure find ERPNext easier; those wanting maximum flexibility prefer Odoo.

5. Maintenance and upgrade policy

Odoo: major version every year (v18, v19, v20...); each supported ~3 years. Upgrading between versions isn't trivial — all custom code must be tested. Odoo.sh (PaaS) automates upgrades but costs money; on-premise you do it yourself. Community upgrades are harder than Enterprise's due to fewer migration tools.

ERPNext: different versioning — continuous patches (v16.x.x), no frequent major versions like Odoo. Upgrading v15→v16 works but needs testing. Frappe framework upgrades sometimes need manual intervention. Pro: very frequent releases — bug fixes weekly. Con: you must track updates constantly, not "set and forget."

Winner: ERPNext for continuous maintenance (frequent but small updates) — Odoo wins on stability within a major version, but cross-version upgrades are harder.

6. Total cost of ownership

License:

• Odoo Community: free (LGPL) but basic features only
• Odoo Enterprise: ~€25–40/user/month (annual billing) + implementation
• ERPNext: completely free (GPL v3) — no license fee

Implementation: both need consultants — ERPNext tends to be slightly cheaper because the community is more open and Python devs are easier to find. Odoo consultants are easier to find but charge more due to demand.

Hosting: both self-host fine — ERPNext slightly lighter (Python + MariaDB + Redis); Odoo is also light but OWL assets are heavier. Both deploy on Docker/VPS.

Approximate 3-year TCO (10-person team, mid-size system):

• Odoo Community: license 0 + implementation ~300–800k THB + infra ~50k/yr
• Odoo Enterprise: license ~200–400k/yr + implementation ~500k–1.5M
• ERPNext: license 0 + implementation ~200–600k + infra ~50k/yr

Cost winner: ERPNext — free license forever, cheaper implementation, more open community. But if you need Enterprise-only features (full accounting, HR payroll), recalculate TCO.

7. AI-generated ERP — the third option

This is the rising question — have AI (Claude, GPT, Cursor) write your own ERP from scratch: schema, business logic, even UI. I've tried several projects; here's the honest take:

What AI does well:

• Basic CRUD + form validation — very fast, usable in 1–2 days
• Simple reports, dashboards, basic workflows
• Custom logic that off-the-shelf ERPs can't do
• Integration with external APIs (payment, shipping, e-commerce)

What AI can't do / risks:

• Legally correct accounting — GL posting rules, tax calculation, financial statements need deep domain knowledge; AI writes plausible-but-wrong logic easily, especially edge cases (credit notes, partial payments, multi-currency)
• Auditable inventory valuation — FIFO/LIFO/average must stay consistent; AI writes logic that looks right but has edge-case bugs
• Multi-company consolidation — intercompany transactions are complex; AI often gets them wrong
• Performance at scale — AI writes working queries but doesn't optimize; big tables get slow silently
• Security & compliance — permission models, audit trails, encryption are "passable" from AI but not robust enough for a real ERP

When AI ERP fits:

• Very small business — just needs order tracking + basic inventory, no deep accounting
• Internal tool — used inside your own company, not a product
• Custom workflows ERPs can't do — e.g., niche booking systems, industry-specific tracking
• Prototype/MVP — test an idea before investing in a real ERP

When AI ERP doesn't fit:

• Business needing legally correct accounting
• Business needing auditability — every transaction traced
• Business that will scale — AI writes fast but is hard to maintain long-term
• Teams with no developers — AI helps write but debugging/maintaining still needs humans

The hybrid approach I recommend: use a proven ERP (ERPNext/Odoo) as the core, then have AI write the custom modules/functions the ERP can't do — e.g., a Frappe app or Odoo module for your niche. You get a reliable core plus the flexibility you need.

8. Summary: which should you pick?

Recommendations by situation

• Multi-channel selling (retail, F&B, e-commerce)→ Odoo — built-in POS + e-commerce + CRM
• Manufacturing / factory→ Odoo — better MRP + shop floor
• B2B business, deep accounting/inventory needs→ ERPNext — stronger accounting + stock ledger, cheaper
• Thai SME on a budget wanting cheap ERP→ ERPNext — free license, cheaper implementation
• Large business needing full suite + enterprise support→ Odoo Enterprise — bigger partner network
• Internal tool, no deep accounting needed→ AI ERP — faster and more precise to build yourself
• Custom workflow ERPs can't do→ Hybrid — ERPNext/Odoo core + AI-written module

9. Final warnings

Whatever you choose, what kills most ERP projects isn't the system — it's:

• Unclear business processes — ERP doesn't fix processes that don't exist
• Dirty data — item master, customer master must be cleaned before migration
• Teams that don't buy in — users refusing to change workflow = system goes unused
• Over-customizing — more customization = harder maintenance and upgrades. Stay on standard processes first
• Counting only license, not TCO — implementation, training, migration, maintenance usually cost several times the license

There's no "best" system — only "best for you." If you can tell me your business process, volume, team size, and budget, I can tell you which fits. If you can't, start with ERPNext — it's cheaper, you can try first, then move to Odoo if needed. Starting cheap and upgrading costs less than starting expensive and downgrading.