SIFEN KUATIA Paraguay 2026: four DNIT waves and why Odoo does not work out of the box

One-minute summary

DNIT moved SIFEN from a voluntary phase to four hard quarterly deadlines in 2026: March 2, June 1, September 1, and December 1. In parallel, every government supplier must issue KUATIA from January 2, no exceptions. If your Asunción-based SME bills PYG 2 billion or more a year and your plan was "we will finish by June," the plan no longer works.

Resolution DNIT N.° 41/2025 closed the voluntary phase of the Sistema Integrado de Facturación Electrónica Nacional. Until 2025, SIFEN was the project for companies that wanted to move early. From January 2026, any business selling to the State — goods, services, consulting, public works — issues electronic documents through SIFEN. After that, four quarterly waves cover the rest of medium and small contributors throughout the year.

Some details get mixed up. KUATIA and KUATIA'i are two distinct products inside SIFEN. KUDE is not the document; it is the representation. The standard l10n_py module in Odoo covers less than half of the validations. If you run an SME of 8 to 50 people in Asunción, you need to know which group you sit in and what must be live on the right Monday of the right month.

• Four SIFEN waves in 2026: Group 15 — March 2, Group 16 — June 1, Group 17 — September 1, Group 18 — December 1. In parallel: January 2 for government suppliers (DNIT resolution 41/2025).
• KUATIA ≠ KUATIA'i. The first is for direct integration through DNIT webservices (mid-size and large taxpayers). The second is a simplified DNIT portal for micro-issuers (cap around 500 documents per year).
• KUDE is not the document. It is the PDF representation with a QR code that the recipient receives. The legal document is the XML signed with an accredited PSC certificate.
• The electronic timbrado takes 5 to 10 business days to register. Start the request on May 25 for Group 16, and you will not issue a single document on June 1.
• l10n_py in Odoo Community does not cover SIFEN. You need custom modules or PAC integration. Without that, compliance is fiction.

The four DNIT waves of 2026: a calendar that no longer moves

Until 2024, Paraguay split tax and customs administration. SET (Subsecretaría de Estado de Tributación) handled taxes, DGA (Dirección Nacional de Aduanas) handled customs. Law N.° 7143/2023 merged them into DNIT (Dirección Nacional de Ingresos Tributarios), a single agency with legal succession and technical infrastructure from both.

Since 2017, SET — and later DNIT — has rolled out SIFEN in cohorts. Each DNIT "group" is an administrative collection of taxpayers built by revenue, activity type, and tax category. Between 2017 and 2025, groups 1 to 14 went live: pilots, large taxpayers, exporters, controlled subsidiaries, and large retailers. 2026 is the mass-onboarding phase for medium and small contributors.

The DNIT calendar for 2026:

Taxpayers see their group inside Marangatu, DNIT's personal portal, by RUC number. From the mandatory date, the old paper timbrado loses validity for standard operations. It is accepted only in contingency mode, and only when SIFEN is unavailable for technical reasons confirmed by DNIT.

More regional context if you run multi-country operations: SUNAT Peru 2026, DIAN Colombia 2026, and SII Chile 2026 follow a similar fixed-calendar logic, each with its own traps.

SIFEN architecture: e-Kuatia, e-Kuatia'i, KUDE, and timbrado

DNIT built SIFEN in two layers to cover both large issuers and micro-businesses. Before you pick a tech stack, get clear on the difference between the products and the concepts.

#1. e-Kuatia: the product for direct integration

• Direct connection through DNIT webservices (SOAP, XML following the SIFEN schema).
• Alternative: integration through a PAC (Proveedor Autorizado de Certificación) — Gosocket, Edicom, Sovos, and similar.
• XML signing with a certificate from an accredited PSC (FirmaPy, COIPSA, and similar).
• The issuer stores KUDE and XML (tax statute-of-limitations period: 5 years).
• Programmable; integrates with any ERP.

#2. e-Kuatia'i: DNIT's simplified portal

• Web form for manual entry.
• Cap around 500 documents per year.
• Signing performed by the platform itself (no need for your own certificate).
• Fits micro-businesses, freelancers, and early-stage SMEs.
• Does not integrate with any ERP.

The most important practical difference: e-Kuatia'i does not connect to Odoo or any other system. If you run Odoo, business logic lives in parallel with manual entry on the DNIT portal. For an SME with 20 documents per day or more, that means duplicate work for every sales rep, permanent accounting gaps, and inevitable tax adjustments on the first reconciliation.

#3. KUDE: the representation, not the document

This is one of the most common confusions. DTE (Documento Tributario Electrónico) is the signed XML — that XML is the legal document. KUDE is the PDF representation of the DTE that gets printed, emailed, or shown on a register. It contains:

• Document data (issuer, recipient, line items, taxes).
• CDC (Código de Control): a unique identifier of 44 characters.
• QR code with a verification link to the DNIT portal.
• Specification-mandated visual elements from DNIT.

The recipient does not have to keep the XML: they work with the KUDE and verify the document independently by scanning the QR against the DNIT portal. The issuer, in contrast, keeps both XML and KUDE for the full statute-of-limitations period.

#4. Registration and prep work before the first KUATIA

Before issuing, you handle four blocks of work:

1. Electronic signature certificate. Bought from an accredited PSC (FirmaPy, COIPSA, etc.). Issuance time: 1 to 3 business days. Cost: PYG 800,000–1,500,000 per year (about USD 110–210).
2. Electronic timbrado request in Marangatu. DNIT processes it in 5 to 10 business days. Critical zone: if you sit in Group 16 and start the request on May 25, your timbrado lands after June 1 — on Monday, you cannot issue a single document.
3. Integration mode decision: direct (webservice + custom code) or through a PAC.
4. Odoo configuration if you use the ERP. This is where the real complexity starts.

Odoo + SIFEN: why l10n_py is not enough

The l10n_py module in standard Odoo (Community and Enterprise) ships a base chart of accounts, VAT rules at 5% / 10%, and RUC fields. That is where coverage ends. For SIFEN specifically, it does not:

• Build XML against the SIFEN schema.
• Sign documents with a certificate.
• Send to the DNIT webservice.
• Generate KUDE with a QR per the official spec.
• Handle post-issuance events (cancellation, void, conformity).

On bare Odoo with l10n_py, SIFEN does not work — assume that from day zero. Three viable paths remain:

For an SME with 100 invoices per month, path 3 is usually the most rational choice on cost and time-to-launch. For businesses with 5,000+ invoices per month, custom development starts to save on PAC fees. More on the Odoo ecosystem in Paraguay: Odoo in Paraguay — SME guide.

Five SME scenarios: where it works and where it breaks

Not all SMEs need the same setup. Five typical profiles in Asunción and surrounding areas.

#1. Craftsman or freelance consultant with 5 to 10 invoices per month

e-Kuatia'i covers everything. No Odoo, no PAC, no certificate needed (the platform signs for you). Total compliance cost per year: around USD 200 (accounting support). Investing in an ERP here is over-engineering.

#2. Retail store or restaurant with 50 to 500 sales per day

e-Kuatia + Odoo POS is the optimal architecture. The hard requirement: no POS session can block a sale when SIFEN is unreachable. You need contingency mode (sale with a provisional timbrado, sync once the connection comes back). Stock Odoo POS does not ship with that. You need custom development or a PAC module that closes the offline scenario.

#3. B2B service with 100 to 500 invoices per month

e-Kuatia + Odoo Invoicing (no POS) + PAC is usually enough. Lower complexity than retail: no real-time pressure, you can batch-emit once a day. This is the most common SME profile in Asunción; migration runs 4 to 6 weeks with average project discipline.

#4. Government supplier, any size

DNIT resolution 41/2025: mandatory from January 2, 2026. Often these are companies that, by group (15 to 18), did not need to rush. A contract with DGA, a municipality, or a ministry makes them immediately obligated. If you won a tender in December 2025 and were not in SIFEN, you could lose the first invoice on the contract — and the advance payment with it.

#5. Business with mixed structure (B2B + retail + export)

The standard template does not apply. You need an architecture session: which Odoo modules (POS, Sales, Inventory, Manufacturing) generate documents, how they connect to SIFEN, how exports get treated (a separate document category). Realistic migration window: 8 to 12 weeks with two test iterations in DNIT sandbox.

Five mistakes that repeat before every wave

#1. "We are in Group 18, December is far away"

By December 1, 2026, you do not need to "connect" — you need to be issuing on SIFEN in a stable way. That means certificate purchased, timbrado registered, ERP integrated, accounting and sales trained, contingency tested. Minimum 6 weeks of clean work. At real SME pace (weekends, fires, vacations), 10 to 14 weeks. For Group 18, "calm prep" starts in September, not in November.

#2. Mixing up physical timbrado and electronic timbrado

The old timbrado for paper invoices does not auto-convert to electronic. Two different processes in Marangatu, with separate requests and separate timelines. Companies that "thought they had everything" find out one week before the deadline that the electronic timbrado was never requested — and the request takes 5 to 10 business days to process. After that, a few days of forced silence with no right to issue.

#3. KUDE without QR or with a malformed CDC

KUDE is not "whatever PDF we drew." DNIT publishes the exact specification: where the QR goes, which fields are mandatory, how the CDC is formatted (44 characters, fixed structure: issuer RUC + document type + emission point + number + security code + year + month). A PDF that fails QR verification is formally valid if SIFEN accepted the XML, but the recipient cannot verify it. Large clients return such KUDEs and ask for a redo.

#4. Ignoring post-issuance events

Once SIFEN accepts the XML, the document has a life cycle. Cancellation: voiding within the regulated window (typically 48 hours after issuance). Inutilización: voiding an unused number. Conformity event: recipient confirmation. Many integrations implement only "send" and skip events. The outcome: gaps between accounting and DNIT records, and automatic adjustments during reconciliation.

#5. Not testing contingency until the first incident

SIFEN goes down. Not often, but it does — once or twice a quarter, between 30 and 120 minutes. If your POS blocks the sale when SIFEN is down, you lose revenue at the register. If your B2B issuer simply queues the send, that is acceptable — but the recovery scenario must be implemented and tested. Many SMEs discover contingency only after the first incident, when the register freezes and the sales team starts writing invoices in Word.

Anonymous case: a Villa Morra importer migrating to Group 16

Anonymized profile: an electrical-equipment importer, 11 employees, revenue around PYG 9 billion a year (~USD 1.4 million), office in Villa Morra, warehouse in Mariano Roque Alonso. Before 2025 they ran on local desktop software and Excel for management accounting. DNIT category: Group 16, deadline June 1, 2026.

What they did between January and April 2026:

1. Document inventory. Two weeks to classify every document type they issued: electronic invoice in 5 variants (B2B, B2C, export, import, autofactura), credit notes, debit notes, and GRE (Guía de Remisión Electrónica). Total: 8 templates.
2. Stack choice. Odoo 17 Community + a custom module l10n_py_sifen_ext (extension on top of OCA) + integration through a local PAC. Direct integration against DNIT was ruled out on spec-maintenance cost.
3. Parallel work. In February they brought up Odoo for warehouse and procurement while invoicing stayed on the old system until late April. That gave breathing room to debug the integration without pressure.
4. Test window in DNIT sandbox. March: 200 test documents, 14 validation errors caught (9 VAT rounding, 3 CDC format, 2 export documents with sender-specific fields).
5. Production cut-over. May 1: switched to SIFEN issuance. One month of dual-system bookkeeping. From June 1: SIFEN only.

"The sandbox saved us. Fourteen errors that would otherwise have become fourteen rejected documents in front of the client."

Outcome at deadline: the company issued through SIFEN for around 95% of documents. The remaining 5% are export operations, where the logic is still being tuned (DNIT accepted a transitional regime). Project budget: USD 14,000 (module, data migration, training, 3 months of post-go-live support).

Next steps: checklist and audit

If you sit in one of Groups 15 to 18, or you work with the State, the next step is not "buy a PAC and connect." It is to audit your current setup: what you issue today, in which document types, what the real load looks like, your ERP infrastructure, which events you handle, how contingency is wired.

There is a 32-point SIFEN 2026 checklist — from ordering the certificate to running the contingency test — plus a 60-day Excel tracker.

For parallel reading:

• Odoo in Paraguay — SME guide
• Odoo audit — 30-minute methodology
• Real Odoo implementation cases across LATAM
• Rescuing Odoo projects that got stuck
• Neighboring regulators: SUNAT Peru 2026, DIAN Colombia 2026, SII Chile 2026

I run Odoo compliance audits and LATAM localizations from Lima, with remote partners in Asunción. If you want a 30-minute conversation about your specific situation, book a slot.

Frequently asked questions