Question 1

How long does onboarding take?

Accepted Answer

Connect one tool, draft one workflow, approve one memory. The first useful action ships in under an hour. Full enterprise rollout — including SAML/OIDC SSO and per-user namespace setup — takes one week with the founders on the line.

Question 2

Where does our data live?

Accepted Answer

Inside your deployment. The substrate runs on your cloud or your hardware — Google Cloud, NVIDIA-backed inference, or your private VPC. No third-party hop. Per-user LibSQL namespace + AES-256-GCM at rest. Supply-chain detail is documented on the Security page.

Question 3

Can we bring our own LLM?

Accepted Answer

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.

Question 4

What hardware is supported?

Accepted Answer

Anything that speaks Modbus, OPC-UA, MQTT, REST, gRPC, serial, or USB-HID. Phones, fixed cameras, drones, rail gantries, spectrometers, harvest arms — UPII negotiates the schema on first contact and signs every frame.

Question 5

How do you handle our existing systems?

Accepted Answer

MCP custom 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.

Question 6

What does pricing look like?

Accepted Answer

Two personal plans: Free, and Pro at US$10/mo with a 7-day free trial — see nexupira.com/pricing for the full comparison. NEXUPIRA is BYOK: you bring your own LLM API key and your provider bills the model usage; we never charge for tokens. Enterprise is custom — sized to the deployment scope and instrument count, not seats, with founders working with each pilot directly.

Question 7

Is NEXUPIRA China-origin-free?

Accepted Answer

Yes. The built-in stack — code dependencies, model providers, cloud regions, hardware suppliers — is documented as zero China-origin. Compliance teams approve without a six-week review. Customers may BYO Chinese-origin components at runtime as a documented exception (e.g., DeepSeek via Custom provider), but the default platform stays clean.

Question 8

Can NEXUPIRA run on-premise or air-gapped?

Accepted Answer

Yes. The full substrate runs on customer-controlled infrastructure — bare metal, private cloud, or fully air-gapped networks. With a local LLM (Llama 3 / Mistral / Qwen via Ollama or vLLM) and on-prem object storage, no traffic ever leaves your perimeter. UPII handles instrument ingestion offline; sync to a control plane is optional.

Question 9

Does NEXUPIRA replace n8n / Make / Workato?

Accepted Answer

For pure SaaS-to-SaaS automation, n8n is enough. NEXUPIRA wins where n8n breaks: precision-instrument ingestion (Modbus, OPC-UA, spectrometers, cameras), running workflows inside your VPC with audit-grade traceability, and treating LLMs as first-class workflow nodes instead of external API calls. Buyers commonly keep n8n for marketing automation and use NEXUPIRA for OT + AI.

Question 10

Who are NEXUPIRA's strategic partners?

Accepted Answer

NEXUPIRA partners with NCIR (National Center for Instrumentation Research, part of NIAR) for precision-instrument and spectral analysis research, and with LOFTY MEMS for precision-instrument and industrial-automation domain expertise. Google Cloud is our platform partner. Partnerships supply deep OT know-how to the platform; NEXUPIRA is an independent company that owns its engine and IP.

Question 11

Who built NEXUPIRA?

Accepted Answer

NEXUPIRA was founded in 2026 and is led by CEO Frank Wu (吳文斌). Frank brings over a decade of hands-on OT experience across precision instruments and industrial automation — medical instruments, motor components, automation equipment development, AGVs, precision fabrication including laser, and image recognition / OCR. Past engagements span HTC (Poland), Quanta Computer, Delta Electronics, Kinik (wafer cutting), Kinpo Group, with research participation on programs at NARI (National Atomic Research Institute). NEXUPIRA channels that decade of OT depth into a next-generation AI iPaaS.

Question 12

Do you train models on customer data?

Accepted Answer

No. Customer data never leaves your deployment to train any provider model. The Workflow Engine's audit log records every LLM call with the provider tag — you see exactly which model touched which data. If you BYO an Ollama / vLLM endpoint, model weights stay entirely under your control.

Question 13

Does NEXUPIRA support Modbus / OPC-UA / MQTT?

Accepted Answer

Yes — all three, plus REST, gRPC, serial, and USB-HID. UPII (Universal Precision Instrument Interface) is NEXUPIRA's industrial ingress layer: it negotiates the protocol on first contact, generates a signed schema, and mounts the instrument as a typed event source. Adding a new factory-floor protocol is a connector update, not a platform rewrite.

Question 14

Can NEXUPIRA do agriculture or strawberry farm workflows?

Accepted Answer

Yes — already live. 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.

Question 15

Can NEXUPIRA do multi-site temp workforce management?

Accepted Answer

Yes — running in chaperoned build. 5 commercial venues, 500+ daily temp workers, 10 supervisors. LINE LIFF is the only worker entry point — face check-in, OCR ID, automated insurance enroll/cancel, site-configurable pay rules, daily payroll. Same Workflow Engine as the strawberry farm — same engine, different persona. Full case study at nexupira.com/customers/workforce.

Common questions

NEXUPIRA subscriptions are in closed beta