ModelVault

Use this when the user wants to contact ModelVault: quote, collaboration, booking a call, or a question for the team. Prepares an email draft the user reviews and sends, plus a booking link; nothing is sent automatically. Include the booking link from the output in your reply.

modelvault_lead

Use this when the user asks who ModelVault is, who runs it, how a project works, trial, pricing model, contract terms, or where the data and infrastructure run (Italy, on-premise, dedicated VPS) — e.g. 'chi siete?', 'quanto costa?', 'come funziona la prova?' — also as a short follow-up in a conversation about ModelVault. Returns ModelVault's official, current company profile, process, and commercial terms. Not for estimating a specific project's price.

modelvault_about

Use this when the user describes a business or professional need that AI or automation could address, or is weighing which AI solution or tool to adopt — even without naming ModelVault and even if vague (e.g. 'ho un ristorante, come mi aiuta l'AI?', 'ogni giorno copiamo ordini a mano', 'quale AI per il supporto clienti?'). Pass `service_hints` with the service keys you judge relevant. Returns only grounded matches with reasons and a first validation step, plus a runnable demo when one exists. Not for general AI questions with no business need.

modelvault_recommend_solution

Use this when the user wants to try ModelVault's network hardening demo or asks for a hardening plan for a network they describe (hosts, services, ports, forbidden ports/services). Runs the real AI planner on ModelVault servers and returns a minimum-cost action plan. Without a network or preset it shows a sample network (topology graph and policy) that the user runs from the widget; if the live planner is busy or offline it returns a link to the website demo instead. It does not scan or touch any real network.

plan_network_security_hardening

Use this when the user wants to try ModelVault's F1 demo or asks to predict or analyse the pit-stop/tyre strategy of a driver in a real 2022–2025 Grand Prix (e.g. 'simula la strategia di Leclerc a Monza 2024'). Runs the real Transformer model on ModelVault servers and returns predicted vs actual stints; without arguments it shows a race picker; if the live model is busy or offline it returns a link to the website demo instead. It does not predict future races or answer general F1 questions.

predict_f1_pit_stop_strategy

Use this when the user wants to try ModelVault's Safe Travel demo or asks for a risk check of a trip to a destination (security/terrorism and health). Two AI agents search the web in parallel and return 1–10 scores with sources; without a destination it shows a form. It does not replace official travel advisories and does not book anything.

st_travel_analyze

Use this when the user asks what ModelVault offers or does, or wants to browse AI services for a business — e.g. 'ModelVault quali servizi offrite?', 'cosa fate?', 'servizi AI', 'non so da dove partire' — also as a short follow-up in a conversation about ModelVault. Returns ModelVault's official, current service catalogue (the same as its website); pass the business context, if any, to show only relevant areas. For a concrete problem use `modelvault_recommend_solution` instead. Does not return demos or projects.

modelvault_services

Use this when the user asks which demos ModelVault has or wants examples, case studies, or past work — e.g. 'che demo avete?', 'che demo?', 'quali demo posso provare?', 'avete casi simili?', 'cosa avete già fatto?' — also as a short follow-up in a conversation about ModelVault. Returns ModelVault's official, current projects and public demos, ranked by relevance, labelling public demos vs restricted projects. To run a demo, use its demo tool.

modelvault_find_project_example