ShenZhen Nebula Rover Technology Limited·深圳市星云智航科技有限公司Hotline400-990-0246Emailperry.xu@airover.net
Technology, security and delivery

One codebase,
two ways to deliver

One language, one repository, one deployment. Browsers never touch the database directly; every business table has row-level security, so even an application bug cannot reach another company's data.

≈208k
lines of TypeScript
148
database migrations, each with a rollback
81
tables with row-level security, 132 policies
764
automated test files
≈80
permission actions, deny by default
≈40
architecture decision records

As of October 2026.

How a request flows

Eight steps behind one question to the bot

01 · Receive

The WeCom long connection receives the @-message and finds the object bound to the extension.

Model integration

Replaceable models, protected keys

One gateway

Chat, streaming, tool calls and embeddings go through one module; Qwen by default, selectable per agent.

Bring your own model

Configure OpenAI-compatible endpoints (e.g. DeepSeek, Zhipu, private vLLM) for chat, embedding, rerank, speech-to-text and text-to-speech; errors fall back to the platform default.

Usage and quotas

Every call records tokens, model and source; usage over the 30-day quota trips a circuit breaker.

Model gateway

On-premises servers hold only a revocable, metered token; real keys never land on them.

SaaS multi-tenancy

One platform for many companies, isolated row by row

One tenant per company

Live

All data carries a company ID; WeCom credentials, model keys and MCP credentials are encrypted per company and invisible to others.

One account, many companies

Live

Membership decides which companies you can enter; the active company decides who you act as; switching goes through a controlled interface.

Tenant lifecycle

Live

Open, suspend (immediately) and re-enable; each tenant has model and storage quotas.

Permission console

In development

Role × page, role × data scope and department × business data in one console.

SaaS and on-premises

SaaS ready to use, or on your own servers

AspectSaaSOn-premises
ServersAIPaddle cloudYour own Ubuntu servers, alongside existing systems
DatabaseShared, with row-level multi-tenant isolationA dedicated database holding only your company
ScopeEverything, including platform operationsPlatform operations removed at build time
Deliverables—Compiled output and configuration only; no source or internal documents
ModelsPlatform keys or your ownThrough the model gateway, real keys stay off your servers; or bring your own
ReleasesMerge → full tests → automatic deployManual approval → build, trim, verify → migrate → atomic switch → automatic acceptance, rollback on failure
BackupsPlatform backupsDaily backups kept for 14 days
One version promotion

Every upgrade follows the same nine steps

01 · Pick

Only commits that passed the full test suite in staging.

Security and permissions

Positions decide the work, roles decide what you can see

Live

Tiered AI autonomy

Reversible actions run directly and can be undone within 12 hours; irreversible ones need the owner's approval.

Live

Security scan on save

Every save is scanned: hard-coded secrets blocked, injection risks fixed.

In development

Positions separate from roles

Positions are the work and drive assignment; roles are permissions: sponsor, owner, member, supervisor.

In development

Department data scopes

Four scopes: all, department and below, department, self.

In development

Approval center

All sharing and grants of AI capabilities are approved in one place, showing who gets access now and who has it already.

In development

KPIs only for those who need them

Work records and KPIs are visible only to the PM, sponsor and admins, never posted to groups; full chat logs are not collected.

Roadmap and progress

Daily loops first, then governance and advanced features

LiveAvailable in production
  • WeCom foundation: org sync, group bots, real @-mentions, pushes
  • Project spaces: auto-created spaces, WeDrive management, inbox filing, file digests
  • PMO basics: group binding, role-based assignment, deadline changes, two-way plan sync, weekly and monthly reports
  • Assistant: build all nine object types, call by name, Word reports, voice and drag-and-drop
  • AI platform: create, edit, scan, approve, grant, extensions
  • Delivery: SaaS and on-premises, model gateway, daily backups and leak checks
In developmentIn staging or in development
  • Checks: due-soon warnings, escalation, status chasing, review logs
  • Rule enforcement: approval matrix, cited notices, daily reports
  • Teams: positions vs roles, AI workers in roles
  • Mobile workbench, chat archive, Office co-editing
  • Governance: permission console, approval center
PlannedConfirmed and scheduled
  • Policy as configuration, health checks, meeting facilitation, issue log, acceptance gates, KPIs
  • Dynamic assignment, proactive unblocking
  • Dependencies and Gantt, plan vs actual, change orders, reports
  • Resource scheduling, margins, clause extraction, retrospectives
  • Platform: more workflow nodes, on-premises with your own database, license binding

Limits we tell you up front

  • On-premises currently uses a dedicated cloud database project; keeping all data inside your network requires self-hosted PostgreSQL, which is planned.
  • Models default to Qwen through the AIPaddle gateway; you can also configure your own OpenAI-compatible endpoint.
  • License machine binding and offline grace periods are planned.

Want to see it running in your business?

A 30-minute online demo built around your industry, including how we would connect to your WeCom and ERP.

Book a demo