Where is the data processed?

It depends on the deployment option and the model. Lattice runs in its cloud, on a VPS or VPC of your organization, or on your own server, which we set up with you as a project. With Na'at and Séeb on your infrastructure or in the Mexico region of our cloud, information is processed in Mexico; with other models, it goes out to the provider shown in the catalog. We tell you in writing, model by model.

The people your organization authorizes. You split teams by area, assign roles and they sign in with Google Workspace, Microsoft Entra ID or any OIDC provider.

The organization's audit log is append-only, exports to CSV, and your organization sets how long it is kept, with a 30-day minimum. Usage is visible by agent, key and model.

What requires approval?

Every message the agent sends is shown in full and only goes out if a person approves it. Risky commands and, for cloud agents, pages that look like payment pages are stopped by automatic detection, which may not catch every case. For everything else, such as filling in and submitting a form or writing from a portal or webmail, your organization can set a policy by organization, team or agent: what is allowed, what asks and what is prohibited. If nobody answers an approval, the action does not run.

What about the models?

You choose which ones each team can use and you see the provider of each one. Sintérgica does not use your data to train models without an explicit contract.

What runs the agents?

Azca, Sintérgica's agent engine. It comes inside Kaná and runs in every deployment option: on the person's computer, in one container per agent in the Lattice cloud, or on your server. It is the piece that calls the model, uses the tools and applies your organization's policy before each action.

Where Lattice runs

Your agents run where you decide, with the models you allow.

Choose our cloud, a server of your organization, a dedicated cloud or your own infrastructure. You can move up when your operation needs it.

Review deployment options See the options

Four deployment options
Policies by organization, team or agent
Activity log

Why it matters

Where your agents run is a business decision

Every operation needs something different. Define it before you start and put it in writing.

Information that must stay inside

Some processes handle data your regulation or internal policy does not allow to be processed outside. For those cases there is a server of your organization or your own infrastructure, with models running right there. If you allow a third-party model, the text the agent sends it goes out to that provider, even if Lattice runs on your server.

Models you allow

Each model in the catalog states its provider and you decide which ones each team may use. With Lattice Na'at and Séeb on your infrastructure or in the Mexico region of our cloud, information is processed in Mexico.

Cost and speed to start

Starting in the Lattice cloud requires no installation. Your own server gives more control, but it is a project with its own timeline.

You hold the key

You decide where it runs, which models are allowed and who enters each space.

Illustration of a specialist next to a server, holding a shield made of nodes

Deployment options

Four ways to run Lattice.

All include unlimited users, teams by area, allowed models per team, approvals, audit log and support in Spanish.

Start today

Lattice cloud

Shared, managed by us

Start with nothing to install. We run the infrastructure, updates and support.

Ideal for: Teams that want their first process running in weeks.

Plan: Teams or Business
Agents included: 5 or 10

As a project · Your server

Your organization's server

A VPS in your organization's name, managed by us

Lattice runs on a virtual private server (VPS) of your organization and we manage it. You decide where information lives without building an IT department.

Ideal for: Companies with several teams that want their data on their own server.

Plan: Business or Enterprise
Agents included: 10 or 25

As a project · More isolation

Dedicated cloud

A private cloud (VPC) just for your organization

Your organization gets its own network in the cloud, separated from other customers, with the same managed operation.

Ideal for: Enterprises with several teams and more agents in production.

Plan: Enterprise
Agents included: 25

As a project · Inside your perimeter

Your own infrastructure

On-premise: platform and engine at your facilities

The platform and the agent engine run on your infrastructure, with separate environments. Models run there too if your project includes inference hardware.

Ideal for: Government and organizations that need to operate inside their perimeter.

Plan: Government and regulated sectors
Agents included: 25 or more

Comparison

The four options side by side.

Each option is quoted with our sales team. The signed proposal sets the contract terms.

Feature	Lattice cloud	Your organization's server	Dedicated cloud	Your own infrastructure
Status	Available today	As a project	As a project	As a project
Plan	Teams or Business	Business or Enterprise	Enterprise	Government and regulated sectors
Who operates it	Sintérgica	Sintérgica, on your server	Sintérgica	Per contract
Processing in Mexico	With Lattice Na'at and Séeb in the Mexico region	With Lattice Na'at and Séeb, on your server	With Lattice Na'at and Séeb in the Mexico region	With Lattice Na'at and Séeb, on your infrastructure
Agents included	5 or 10	10 or 25	25	25 or more
Custom specialized model	Per project	Per project	Per project	Per project

In every option

What is included

Unlimited users: you pay for agents and what they use
Teams by area and allowed models per team
Approval of every message the agent sends, plus a per-tool policy: allow, ask or prohibit
Append-only audit log, exportable to CSV, with usage by agent and model
Catalog connectors
Usage alerts and spending caps
Support in Spanish

Sectors

Where choosing well matters most

Financial

CNBV, credit, AML

Government

Runs on your own infrastructure

Healthcare

Health regulation

Legal and notarial

Case files and contracts

Energy

Regulatory reports

Logistics and foreign trade

Customs and files

Your data does not train our models

Sintérgica does not use your data to train models without an explicit contract. With third-party models, each provider's terms also apply, and you can see them in the catalog. With masking turned on in your policy, email, phone, RFC, CURP and card numbers are replaced with a label before they reach the model. Your data and your configuration are yours and are delivered in open formats.

For IT and security

What your IT team will ask

Where is the data processed?: It depends on the deployment option and the model. Lattice runs in its cloud, on a VPS or VPC of your organization, or on your own server, which we set up with you as a project. With Na'at and Séeb on your infrastructure or in the Mexico region of our cloud, information is processed in Mexico; with other models, it goes out to the provider shown in the catalog. We tell you in writing, model by model.
Who has access?: The people your organization authorizes. You split teams by area, assign roles and they sign in with Google Workspace, Microsoft Entra ID or any OIDC provider.
What gets logged?: The organization's audit log is append-only, exports to CSV, and your organization sets how long it is kept, with a 30-day minimum. Usage is visible by agent, key and model.
What requires approval?: Every message the agent sends is shown in full and only goes out if a person approves it. Risky commands and, for cloud agents, pages that look like payment pages are stopped by automatic detection, which may not catch every case. For everything else, such as filling in and submitting a form or writing from a portal or webmail, your organization can set a policy by organization, team or agent: what is allowed, what asks and what is prohibited. If nobody answers an approval, the action does not run.
What about the models?: You choose which ones each team can use and you see the provider of each one. Sintérgica does not use your data to train models without an explicit contract.
What runs the agents?: Azca, Sintérgica's agent engine. It comes inside Kaná and runs in every deployment option: on the person's computer, in one container per agent in the Lattice cloud, or on your server. It is the piece that calls the model, uses the tools and applies your organization's policy before each action.

Next step

Choose the option with your security team

In 45 minutes we tell you in writing which option fits, with which models and what it costs. The session is free.

Review deployment options

Free
Written answer
Model by model