Documentation / Deployment modes & private deployment

Deployment modes & private deployment

ClawTouch ships in three deployment shapes, picked based on how large the rollout is, what your network looks like, and how sensitive the data is. This page walks through what each one includes, what it requires, and roughly how a rollout unfolds — written so your IT team can sketch the path to production.

01 · Three deployment shapes

02 · Standard mode

Best fit: anywhere from a few seats to a few dozen, with the operations team running things day to day from the WeChat mini-program. This is ClawTouch's default shape.

What you get:

• Desktop client (Windows application) × N machines
• HID device × N (one per machine)
• The WeChat mini-program for your operations team
• Our built-in AI model — no API key to provision
• Our backend handles accounts, subscriptions, device pairing, tickets, and customer support

A typical rollout looks like:

1. Sign the contract bundle: MSA + DPA + SLA
2. We ship the HID devices
3. You install the desktop client on each Windows machine
4. Operators scan a code in the mini-program to pair devices and activate accounts
5. The operations team configures task templates, the persona library, and managed-mode policies in the mini-program

03 · Client & hardware requirements

Desktop client

HID device

• Manufactured by us in batches, with a unique serial number burned in at the factory
• One device per machine; the physical identity is bound to its software license
• Defective units are replaced under the SLA
• Firmware updates over the air require explicit customer authorization

04 · Fleet deployment (50+ devices)

For deployments of 50 devices and up, we ship a dedicated desktop admin console. It either replaces the mini-program or sits alongside it:

• Instead of managing devices one phone at a time, IT sees the whole fleet from a Windows desktop
• Push task templates, the persona library, and managed-mode policies in batches
• Multi-user roles and permission tiers
• Full operational dashboards with export
• API / webhook integration with your existing CRM, ticketing, or OA systems (built on request)

The desktop admin console isn't only for offline deployments — it's also available for online deployments. Exact capabilities are scoped per customer; details land in commercial discussions and the SOW.

05 · Offline private deployment

Best fit: customers in finance, healthcare, government, defense, or anywhere else data simply cannot leave the company perimeter. This is a custom add-on, scoped to each customer's size and use case.

What it looks like:

• The desktop client runs on customer machines
• The server-side components ship inside the same package and run on the customer's internal network
• The LLM runs locally or on the customer's internal network — options include your own model, a fine-tuned open-source model, or exclusive weights we build for you
• The desktop admin console connects to the internal network only; no internet dependency

What we provide:

• On-site deployment support
• Local LLM selection and fine-tuning
• Offline upgrade packages with documented rollout procedures
• Training for your IT team

Specific technical details, hardware spec, offline license-activation mechanism, and SLA terms are all scoped per customer. For a needs assessment, write to support@tinqiao.com.

06 · Network topology

Mode A · Standard
─────────────────────────────────────────
   Desktop client ──HTTPS──→ LLM API
        │
        ├──HTTPS──→ ClawTouch backend
        │           (accounts / subs / devices / tickets)
        │                 ↑
   ┌────┴────┐            │
   │   HID   │       Mini-program
   └─────────┘


Mode B · Fleet (50+ devices)
─────────────────────────────────────────
   Desktop client × N ──HTTPS──→ LLM API
        │
        ├──HTTPS──→ ClawTouch backend
        │                 ↑
   ┌────┴────┐            │
   │   HID   │      Desktop admin
   └─────────┘      console (on-prem IT)


Mode C · Offline private deployment
─────────────────────────────────────────
        ┌── Customer network (air-gapped) ──┐
        │                                    │
  Desktop client × N ──→ Customer-hosted LLM
        │
        ├──→ Embedded server ←─ Desktop admin
        │
   ┌────┴────┐
   │   HID   │
   └─────────┘

07 · Typical rollout timeline

Offline private deployments add another 2–4 weeks for local LLM setup and on-site support.