Questions.
Answers.

Mobox is an independent AI engineering firm based in Assago (MI), Italy. We design Artificial Intelligence, Cybersecurity, Big Data and Automation solutions for companies that want to grow in a measurable way.

Yes. We work with SMBs, mid-market and enterprise. The engagement model changes: for SMBs we start from compact packages and MVPs, for enterprises we run multi-year programs with dedicated governance.

Manufacturing, retail, finance, healthcare, logistics, energy and public administration. Our approach is vertical: domain first, technology second.

Mobox S.r.l. — Via Roggia Bartolomea, 7 — 20057 Assago (MI), Italy. We operate across Italy and the EU, also remotely.

Artificial Intelligence (LLMs, agents, RAG, computer vision), Cybersecurity (assessment, SOC, GDPR/NIS2), Big Data & Analytics, Process Automation (RPA, workflow), Cloud & DevOps.

Services are custom projects tailored to the client. Products — ArtUP, Fencity, Nexus — are proprietary platforms ready to use, configurable and integrable with existing systems.

Yes. ArtUP, Fencity and Nexus expose REST APIs, webhooks and connectors for the major ERP, CRM, IAM and data warehouse systems on the market.

Both. We start from a discovery phase (workshops, audit, roadmap) and continue, if it makes sense for the client, with design, implementation, change management and operational support.

An AI MVP starts from 6–10 weeks. An enterprise program (e.g. a data platform or SOC) develops over 6–18 months with incremental releases every 2–4 weeks.

Fixed-price on closed scopes, time & material on evolutionary programs, monthly retainer for ongoing management. Every proposal includes milestones, deliverables and acceptance criteria.

Yes. Scrum or Kanban depending on client maturity, with bi-weekly demos, shared backlog and outcome (not just output) metrics.

We offer support SLAs (8x5 or 24x7), evolutive maintenance, proactive monitoring and training paths for internal teams.

Both. We choose based on cost, latency, privacy and quality: OpenAI, Anthropic, Google and Mistral on the proprietary side; Llama, Qwen and vertical models on the open-source side, also self-hosted.

No. We only work with providers and configurations that guarantee no-training on client data. Where required, on-premise deployment or in a European private cloud.

Retrieval-Augmented Generation: the model answers based on your documents, not only on its general knowledge. It's useful for internal knowledge bases, customer support, compliance and document search.

We define product KPIs upfront (e.g. resolution rate, accuracy, latency) and set up automatic evaluation pipelines, with test datasets and production monitoring.

Yes. All projects are designed privacy-by-design: DPIA when necessary, DPA contracts, data minimization and access logging.

We offer dedicated assessments to identify gaps and priorities. For the AI Act we classify systems by risk level and implement transparency, human oversight and technical documentation requirements.

Yes, by default. We work with European regions of the main cloud providers and, on request, with full-EU or on-premise deployments.

We operate with controls aligned to ISO/IEC 27001 and NIST CSF. Specific details (audits, attestations) are shared under NDA with clients.

Write to support@mobox.it or use the form on the Contact page. We'll get back to you within 24 working hours to schedule a free discovery call.

Yes. A 30-minute call to understand the context, the goals and whether we're really the right partner for you. No commitment.

Yes. We have a partner program for system integrators, software houses and consulting firms that want to integrate AI and automation into their own projects.

Visit the Work with us page. We look for AI, data, security and engineering profiles — also juniors with strong technical curiosity.