LOCAL AND HYBRID AI · BUILT IN ITALY

Your organisation’s AI.
Within your perimeter.

Documents, models and agents in one operational environment. Bring AI into business and institutional workflows, choosing which tasks stay local and which services to connect.

  • Selected data and sources
  • Configurable autonomy
  • Mac Apple Silicon

SOLUTIONS

One platform. Different operational needs.

A team working together01

Businesses and teams

Turn procedures, documents and recurring tasks into reusable working knowledge. Start with a specific process.

Explore
A dedicated local workstation02

Institutions and public bodies

Organise internal knowledge and prepare drafts and summaries with accessible sources, clear responsibilities and human review.

Explore
Illustration of a contemporary campus03

Buildings and infrastructure

Evaluate a dedicated AI node for a building, campus or operational site. A pilot project for developers, operators and integrators.

Explore

T-CHAT 4.6.5

From conversation to real work.

01

Documents become usable knowledge

Second Brain brings selected sources together, reads Markdown and refreshes context when documents change. New additions proposed by AI require your confirmation.

Discover the feature
02

Agents that work on files

Projects, Workspace and local applications, including Python and databases. File writes are verified and the editor stays in sync while protecting unsaved edits.

Discover the feature
03

An AI team with clear rules

Personalities, multi-agent meetings, missions and automations. Permissions, confirmations and Anima follow the settings you enable.

Discover the feature
04

Models suited to the machine

Ollama, MLX and llama.cpp, with API services when enabled. Memory management adapts to hardware and coordinates models during meetings too.

Discover the feature
05

Connected knowledge and tools

Email, calendars and supported accounts join the workflow through explicit permissions. Sources destined for Second Brain are selected before import.

Discover the feature
06

Clearer local control

Companion manages runtimes and models with Hugging Face download progress, data amounts and error details. Nearby extends access to devices on the configured network.

Discover the feature

LOCAL · HYBRID

Choose where the work happens.

Local AI

Models and selected knowledge on your infrastructure. Useful for internal documents and tasks that should not depend on a remote model provider.

Connected services

Enable external models and integrations when the task requires them. Decide which data may leave the local environment and who can authorise actions.

Local features can run without Internet when models, runtimes and documents are already available and the licence is valid. The offline period depends on the licence; renewals, power and recovery need planning. Email, APIs and other online integrations require connectivity.

PILOT PROJECT

Start small. Measure what matters.

01

Define the scope

Choose a workflow, approved documents, a responsible person and an acceptable level of error.

02

Test real work

Measure answer quality, response time, memory use and behaviour when sources or connectivity change.

03

Plan the rollout

Agree on access, training, maintenance, licence renewals and recovery before expanding access.

T-CHAT JOURNAL

Practical decisions for your AI project.

01

Your first business AI pilot

A practical method for comparing results, costs and responsibilities.

Read the guide
02

An AI node for a building

Documents, workstations and infrastructure: what to validate before installation.

Read the guide
03

Offline information in shelters

Power, current sources and operational limits in a continuity plan.

Read the guide

YOUR NEXT STEP

Bring a real use case. Let’s build the pilot.

Tell us what your organisation needs to achieve. We will discuss workflows, infrastructure and a practical demonstration.

business@mytichat.com

Required fields except phone.

FAQ

Before you start.

Does T-Chat work without Internet access?

Local features can run without Internet when models, runtimes and documents are already available and the licence is valid. The offline period depends on the licence; renewals, power and recovery need planning. Email, APIs and other online integrations require connectivity.

Which hardware is supported today?

The current edition supports Mac Apple Silicon. Memory, model and workload should be sized for the actual use case. NVIDIA and AMD versions are in development, with availability not yet announced.

Can several people use it?

Nearby provides access for authorised devices on the configured network. Concurrent users, models, permissions and response times must be tested during the pilot: a single machine does not have unlimited capacity.

Is this already an emergency or shelter system?

We propose evaluating a local information node for documentation and staff support. It is not a certified emergency system and does not replace official communications, procedures, responders or qualified professionals.

Is data always processed locally?

That depends on configuration. Local models run on the machine; enabled API services and integrations may transmit the data required by the selected service. Define the perimeter before introducing sensitive documents.