Hardware and installation
The proposal will specify configuration, delivery, installation, warranty and operating responsibilities. It will also state ownership, return or any buyout option for the machine.
Private AI · On-prem
Your artificial intelligence. In your company.
Workspace, documenti e agenti AI nella tua infrastruttura, con modelli locali compatibili. L’offerta è in pre-lancio: configurazione e disponibilità si verificano insieme.
Pre-lancio: disponibilità e configurazione da verificare.
Piattaforma UTOMAI · Private AI On-prem
La piattaforma AI.
In your infrastructure.
UTOMAI porta workspace, knowledge e agenti configurabili nella tua azienda. La macchina dedicata è la modalità di distribuzione: Syncronika cura installazione, configurazione e servizi di gestione previsti nell’offerta.
Modelli, compatibilità, accessi e perimetro operativo vengono verificati e riportati nella proposta prima dell’adesione.
- Workspace, knowledge e consultazione documentale
- Configurable agents and compatible local models
- Users, permissions and access through the company network or VPN
- Updates, monitoring and assistance according to the contract
UTOMAI On-premLocal modelsYour knowledgeYour agents
Inside your company network
Concept illustration. Hardware selected after assessment.We assess
Team, data and processes, to check compatibility and sizing.
We configure
Appliance, platform and scope defined in the confirmed proposal.
We support you
Activation and managed services according to contractual responsibilities and SLAs.
Pre-launch terms
First we define the hardware, licence and service.
A price will follow the technical assessment: the machine, compatible models, installation and level of management vary by project.
Platform and management
The agreed scope will describe the UTOMAI licence, updates, monitoring, support and service levels. Integrations and custom work will be assessed separately.
Term and end of contract
The proposal will clarify the term, termination, renewal and what happens to the machine at expiry, including any buyout and continued service.
The benefits of Private AI
Company knowledge. Everyday usefulness.
Data sovereignty
More control over data, models and access in your infrastructure, with declared external dependencies and agreed data flows.
Models running locally
Local inference does not require sending data to external LLM providers. Compatible open-weight models are used.
Condizioni definite in proposta
La proposta chiarirà durata, proprietà o restituzione della macchina, eventuale riscatto, rinnovo e servizi successivi.
Company knowledge
Ask questions of internal documents and find the sources behind an answer, with access defined for your team.
Configurable agents
Give agents a task, relevant knowledge, permissions and human review steps.
Connections to your systems
Custom ERP, CRM and database integrations are assessed and quoted separately.
Dedicated sizing
Models, concurrent requests and response times guide the configuration choice, without capacity guarantees in advance.
Professional support
Updates, monitoring, backups and support within the agreed scope.
Local model candidates
Open-weight models, chosen for your work.
Mistral, Qwen, Meta Llama, DeepSeek e GLM di Z.ai sono famiglie da valutare. I modelli effettivamente installabili su ogni configurazione dipendono dalla verifica tecnica, hardware e delle licenze.
Mistral AI
Version, licence, size and local compatibility are confirmed for the selected configuration. A family name does not guarantee every model or capability.
Qwen
Version, licence, size and local compatibility are confirmed for the selected configuration. A family name does not guarantee every model or capability.
Meta Llama
Version, licence, size and local compatibility are confirmed for the selected configuration. A family name does not guarantee every model or capability.
DeepSeek
Version, licence, size and local compatibility are confirmed for the selected configuration. A family name does not guarantee every model or capability.
GLM · Z.ai
Version, licence, size and local compatibility are confirmed for the selected configuration. A family name does not guarantee every model or capability.
The choice depends on quality, documents, expected workload and hardware. External European or non-European frontier models can be assessed on request, with data flows and costs separate from local inference.
One platform. Your choice of infrastructure.
In the cloud.
Or in your company.
Workspace, knowledge, agents and automations. Choose the deployment around your data, your team and how you want to operate.
Choose your AI deployment.
| Your requirement | Dedicated Cloud | On-premise |
|---|---|---|
| Infrastructure | Dedicated cloud servers in Italy; European options on request | Dedicated appliance in your infrastructure |
| AI models | Open-weight models; European or non-European frontier providers on request | Compatible local open-weight models; optional external providers |
| Knowledge | Stored in the configured cloud environment | Documents and indexes can remain in your infrastructure |
| Access | Online access for your team | Company network or VPN, as configured |
| Condizioni | Perimetro, servizi e condizioni definiti nella proposta | Offerta in pre-lancio; perimetro, servizi e condizioni da definire nella proposta |
| Manage | Cloud service and agreed support | Updates, monitoring, backups and support within the agreed scope |
Available features depend on the validated configuration. External integrations, telemetry, backups or remote assistance may involve cloud services and related data flows.
Dedicated Cloud
A clear location. A deliberate model choice.
Italy first. Europe on request.
Private AI runs on dedicated cloud servers in Italy, with European locations assessable on request. Storage, model execution, backups and operating access are confirmed separately in the quotation.
Open-weight as the foundation.
The planned catalogue uses selected open-weight models. External frontier models from European or non-European providers can be enabled on request, with agreed destinations, terms and costs.
Continuity within agreed boundaries.
Failover policies use approved alternative endpoints, respecting the agreed geography and data restrictions. Availability, recovery arrangements and SLA are defined for the configuration; uninterrupted service is not guaranteed.
An Italian workspace does not make every external model route Italian. Model origin, hosting location and inference destination are distinct; the permitted data flows are confirmed before activation.
Applicazioni per le aziende
Confidential information. Practical workflows.
Professional & consulting firms
Answers from approved contracts, files and procedures, with document access defined for each team.
Manufacturing & engineering firms
Support in consulting manuals, quality documents and technical knowledge, with references to sources.
Companies with internal business systems
Knowledge and agents can connect to ERP, CRM and databases, with permissions and action review. Custom integrations are extra.
Illustrative scenarios to validate against your company’s documents and systems. Workflows using external services may require connectivity and additional costs.
Before you start
On-premise private AI, in practice.
Control over data locations, models and access, with declared responsibilities and external dependencies. Local inference can avoid sending inputs to external LLM providers for that processing; external backups, remote access and integrations depend on the configuration. This does not mean total isolation or automatic compliance.
Enabled users and simultaneous AI requests are sized separately. Model, context, concurrency and performance are defined after technical validation. The website does not guarantee a number of concurrent requests or unlimited capacity.
The proposal will specify the software, features and models planned for the configuration after availability and compatibility checks. Custom integrations and workflows are not automatically included.
Fully local functions can operate without mandatory cloud dependencies when all required components are configured and validated locally. External backups, remote access and cloud connectors may require connectivity and agreed data flows.
Before acceptance, the proposal must state whether the machine remains the provider’s property or transfers to the customer, whether return or buyout is available, and the options for renewal, replacement and service after the term. A recurring fee alone does not imply transfer of ownership.
Bring AI into your company.
Let’s discuss your team, documents and first useful application. We assess configuration, installation and terms for your company.