Common questions
Fifteen common questions. Direct answers — no hedging, no SDR script. Want one not on the list? Email hello@nexupira.com.
01 How long does onboarding take?
Connect one tool, draft one workflow, approve one memory. The first useful action ships in under an hour. Full enterprise rollout, including per-account setup, takes about a week with the founders on the line. SAML/OIDC SSO is planned.02 Where does our data live?
Today NEXUPIRA runs as our cloud service. Each account has its own isolated database schema, and sensitive data such as credentials and document contents is encrypted. Platform AI requests go to the model providers listed in our Privacy Policy. Deployment in your own cloud or VPC is planned for Enterprise.03 Can we bring our own LLM?
Yes. The Custom provider slot accepts any OpenAI-compatible endpoint at configuration time — your choice, your endpoint, your audit log. The provider tag travels with every call.04 What hardware is supported?
Live today: sensors that send readings over REST (HTTPS/JSON), and phones and field cameras that send photos. Planned: direct industrial protocols (Modbus, OPC-UA, MQTT, gRPC, serial, USB-HID) through UPII, which will negotiate the schema on first contact and sign every frame. Those adapters are not available yet.05 How do you handle our existing systems?
On Pro and Max, your own remote MCP servers connect your ERP, PLM, CRM, and internal tools. Workflow Engine triggers on Event Bus subscriptions. We coexist with what you already run — we are not a rip-and-replace.06 What does pricing look like?
Personal plans: a free trial, Pro at US$15/month and Max at US$100/month (priced in US dollars; subscribe in the app under Settings → Plans & Billing; every region pays in US dollars); see nexupira.com/pricing for the full comparison. Every plan includes platform AI credits, so you do not need your own key; on Pro and above you can use your own model account instead, and the provider bills you directly. Enterprise is quoted by deployment scope and needs.07 Is NEXUPIRA China-origin-free?
Yes. The built-in stack — code dependencies, model providers, cloud regions, hardware suppliers — is documented as zero China-origin. Customers may BYO Chinese-origin components at runtime as a documented exception (e.g., DeepSeek via Custom provider), but the default platform stays clean.08 Can NEXUPIRA run on-premise or air-gapped?
Not as a product you can install today. Today NEXUPIRA runs as a cloud service. Deployment on customer-controlled infrastructure — private cloud, your own servers or an air-gapped network, with a local model through Ollama or vLLM — is planned for Enterprise and scoped per project. Tell us your requirements and we will say plainly what is possible and when.09 Does NEXUPIRA replace n8n / Make / Workato?
For pure SaaS-to-SaaS automation, n8n is enough. NEXUPIRA is built for where n8n breaks: field data ingestion (sensor readings and camera photos today; Modbus and OPC-UA adapters are planned) and treating LLMs as first-class workflow nodes instead of external API calls. Running workflows inside your VPC is planned for Enterprise.10 Who does NEXUPIRA work with?
NEXUPIRA works with the National Center for Instrumentation Research (NCIR), part of NIAR (National Institutes of Applied Research), on precision-instrument and spectral-analysis research. NEXUPIRA is built and operated by Lofty Mems Co., Ltd. (嶸惠有限公司), Taiwan.11 Who built NEXUPIRA?
NEXUPIRA was founded in 2026 and is built and operated by Lofty Mems Co., Ltd. Jo-Tung Wu is founder and CEO, Tim J is CTO, Esther Chang is CMO, and Frank Wu, who started Lofty Mems in 2013 and has over a decade in precision instruments and automation, is an advisor. Meet the team at nexupira.com/about.12 Do you train models on customer data?
No. Customer content is never used to train AI models. Every decryption of a stored AI key or connection token is written to the audit log with the provider name. If you bring your own Ollama / vLLM endpoint, the model weights stay under your control.13 Does NEXUPIRA support Modbus / OPC-UA / MQTT?
Not yet: all three are planned. Today, sensor readings and camera photos come in over REST (HTTPS/JSON) and can trigger workflows. UPII (Universal Precision Instrument Interface) is the planned industrial ingress layer: it will negotiate the protocol on first contact, generate a signed schema and mount the instrument as a typed event source, so adding a factory-floor protocol becomes a connector update, not a platform rewrite.14 Can NEXUPIRA do agriculture or strawberry farm workflows?
Yes — pilot completed. Shi Jian Strawberry farm (Xinshe, Taiwan) replaced a daily 2-hour manual disease patrol with one phone photo: YOLO + Vision LLM identifies the disease and location in under 10 seconds. Currently extending with NCIR for non-destructive spectral analysis (sweetness, pesticide residue). Full case study at nexupira.com/customers/strawberry.15 Can NEXUPIRA do multi-site temp workforce management?
It is a chaperoned build still in progress. 5 commercial venues, 500+ daily temp workers, 10 supervisors. Face recognition and ID OCR verification are done; LINE check-in, automated insurance enroll/cancel, site-configurable pay rules and remittance reconciliation are being built. Same Workflow Engine as the strawberry farm. Full case study at nexupira.com/customers/workforce.