Lattice Na'at · Sintérgica Labs

A model that understands the context of Mexico.

Sintérgica Labs specializes it with a corpus of Mexican Spanish and Mexican regulation. It is released on November 13, 2026, on Lattice and Hugging Face.

See whether Lattice Na'at fits your process See what we are publishing

Curated Mexican corpus
Release Nov 13, 2026
On Lattice and Hugging Face
Release · November 13, 2026

Why Lattice Na'at exists

The WEIRD bias: the further a culture is from the United States, the worse a global model represents it.

r = −0.70

Correlation between bias and cultural distance

Atari et al., Harvard 2023

≈15%

Share of humanity represented as WEIRD

Behavioral science

Countries compared in the study

94,278 people surveyed

In 2023, Harvard researchers (Atari, Xue, Park, Blasi & Henrich) compared GPT's answers with data from 94,278 people in 65 countries. They found a correlation of r = −0.70 between a country's cultural distance from the United States and the model's ability to represent its people.

WEIRD (Western, Educated, Industrialized, Rich, Democratic) describes the societies that produce most of the content used to train AI models. They make up about 15% of humanity.

Mexico (with legal frameworks based on codified civil law, more collectivist forms of social organization and a regulatory system entirely its own) is systematically under-represented in models trained predominantly on Anglo-Saxon data.

The result is practical: a global model reviewing a Mexican contract may apply Common Law logic.

Lattice Na'at is trained to narrow that gap.

Read the research on WEIRD bias

What we are publishing

Lattice Na'at and the first Séeb vertical, on November 13.

Each model ships with its corpus, its date and its evaluation. Legal and Financial will be announced once their evaluations are ready.

Lattice Na'at

Release · Nov 13, 2026

Specialty: Mexican Spanish and Mexican context
How it is used: Credits in your plan
Where: Lattice · Hugging Face

What for

Drafting, summarizing and reasoning with Mexico's vocabulary, regulation and formats.

Lattice Séeb Logistics

Early access · Nov 13, 2026

Specialty: SAT, Customs Law, TIGIE, 3PL
How it is used: Credits in your plan
Where: Lattice · pilot clients

What for

Looking up customs and foreign trade regulation with the right legal grounding.

Lattice Séeb Legal

Coming soon

Specialty: Civil and commercial legislation, SCJN, amparo, DOF
How it is used: Credits in your plan
Where: Lattice

What for

Comparing contracts, summarizing case files and tracking the DOF.

Lattice Séeb Financial

Coming soon

Specialty: CNBV, CNSF, SAT and CFDI, KYC, AML and PLD
How it is used: Credits in your plan
Where: Lattice

What for

Preparing files and looking up the sector's regulatory provisions.

In Lattice, models are used with the credits included in your plan.

What sets Lattice Na'at apart

Trained for the context that global models ignore.

Lattice Na'at is a model trained to reason outside WEIRD bias, with deep knowledge of Mexico.

01 · Na'at

Mexican Spanish as the source language

Specialized with text written in Mexican Spanish, not translated from English: idioms, legal and technical terminology, institutional formats.

02 · Na'at

Local regulation and context

The corpus includes Mexican regulation and documentation curated by our lab. What is ours is the corpus, the training and the evaluation.

03 · Na'at

Public model card

When it is published on Hugging Face, Na'at will have a model card with its size, base, license and evaluation, so anyone can check what we claim.

04 · Na'at

Processing in Mexico

With Na'at on your infrastructure or in the Mexico region of our cloud, your information is processed in Mexico.

05 · Na'at

Included in Lattice

It is used with the credits in your Lattice plan, alongside the rest of the model catalog, with no other vendor to hire.

06 · Na'at

Science behind it

It comes from Sintérgica Labs, which publishes peer-reviewed research, maintains an open Mexican dataset and studies the cultural bias of models.

Lattice Na'at on the platform

From the lab to your team's work.

The lab researches and creates the models, Lattice manages them and your team does the work in Kaná, its desktop app.

Sintérgica Labs
Research
Curates the corpus, then specializes and evaluates the models.
Lattice Na'at and Séeb
Models
Lattice Na'at for Mexican context; Séeb for each sector's regulation.
Lattice
Agent platform
Where agents, allowed models, approvals and the audit log are managed, and the API is accessed.
Kaná
The desktop app
Where your team does the work, on Mac and Windows.

Track record

From research to the model's release.

This is how we reached November 13, 2026.

Phase 01
2024
Completed
Sintérgica Labs is born
We saw that global models did not understand Mexico's regulatory, linguistic and operational context, and we founded the lab.
Phase 02
2025
Completed
First datasets
First Mexican datasets and first products in operation with clients.
Phase 03
2026
Completed
Peer-reviewed publication
Paper at the Mexican Congress of Artificial Intelligence (COMIA 2026, Springer), with the Tecnológico Nacional de México, Veracruz campus.
Phase 04
Nov 13, 2026
Current
Lattice Na'at release
Lattice Na'at on Lattice and Hugging Face; Séeb Logistics in early access with pilot clients.

Access and use

Three ways to work with Lattice Na'at.

On Lattice

From November 13, 2026, alongside the rest of the model catalog, with the credits included in your Lattice plan.

Discover Lattice

On Hugging Face

Na'at is published with its model card: size, base, license and evaluation. You can already check our lab's public dataset today.

View on Hugging Face

With Sintérgica Labs

Research projects, published or confidential, research residencies, and data donation or exchange for scientific purposes.

Discover Labs

FAQ

Frequently asked questions

Na'at will be published on November 13, 2026, at the Sintérgica AI Summit in Veracruz. From that day it is available on Lattice and Hugging Face.

Na'at is a fine-tune made by our lab on an open base, with Mexican and regulatory corpora curated by us. What is ours is the corpus, the training and the evaluation. The Hugging Face model card details size, base, license and evaluation.

In Lattice, Na'at is used with the credits included in your plan. If they run out, extra usage is deducted from your account balance. Once it is published on Hugging Face, you will also be able to download it.

With Na'at on your infrastructure or in the Mexico region of our cloud, your information is processed in Mexico. With other models in the Lattice catalog, each one states its provider and you decide which are allowed. If you turn on masking in your policy, email, phone, RFC, CURP and card numbers are replaced with a label before they reach the model. For other sensitive data, anonymization is defined in the project.

Na'at does not compete with global models on general tasks in English. It is specialized for the Mexican context: Mexican Spanish, local regulation and local formats. We publish its evaluation, with its method, on the model card.

Séeb are our sector-specialized models. Séeb Logistics enters early access on November 13, 2026; Séeb Legal and Séeb Financial are coming soon.

Collaborate on the research

Let's build data and models that understand Mexico.

If your organization, university or institution wants to contribute to the corpus or the evaluation, or do research with us, we will tell you how to take part.

Propose a collaboration with Labs

Public model card on Hugging Face
On your infrastructure, as a project
Public Constitution