Why TianKey Cloud

Not just another chatbot — but intake, handover, agent collaboration and analytics done right in one platform.

80%+Repetitive questions handled by AI
24/7Always online, no night gaps
10+Channels unified in one desk
¥1,280Standard plan from / year
CORE ADVANTAGES

Eight core advantages

From answer quality to compliant handover, from omnichannel intake to analytics — covering the whole service chain.

🤖

Dual AI engine

Knowledge base retrieval plus LLM generation, with BERT intent recognition and automatic similar-question learning — accurate, traceable answers that improve over time.

  • Industry knowledge templates, live in a day
  • Unanswered questions collected and re-learned
  • Traceable answers, fully controllable
🔗

Omnichannel intake

One dashboard for web, WeChat Official Account, Mini Program, WeCom, Douyin and voice — every message lands in one conversation workspace.

  • One line of JS for the website
  • One-click WeChat binding
  • Automatic Douyin DM replies
⚖️

Compliant human handover

Aligned with GB/T 47746-2026: negative sentiment, explicit requests and exhausted quota trigger handover automatically, with reasons tagged and full audit trail.

  • Configurable handover rules
  • Automatic reason tagging
  • End-to-end audit logging
🪄

Visual flow orchestration v2

Canvas-based multi-turn flows with conditions, API calls, approval steps, scheduled triggers and handover nodes — complex logic without code.

  • WYSIWYG with instant debugging
  • API nodes connect to your systems
  • Approval nodes for human-in-the-loop
🎭

Multi-bot personalization

Route by channel, keyword or customer group to dedicated bots with isolated knowledge bases and brand voice.

  • Channel / keyword routing
  • Isolated knowledge per bot
  • Custom persona and tone
📊

BI analytics

Conversations, response time, CSAT, bot hit rate and agent KPIs in one view, with daily/weekly/monthly reports pushed automatically.

  • Real-time key metrics
  • Hot topics and gap analysis
  • Scheduled report delivery
🛡️

Security & tenant isolation

Tenant-level data isolation, HTTPS encryption, sensitive word filtering, fine-grained roles and audit logs protecting business and customer data.

  • Tenant-level isolation
  • Automatic sensitive word interception
  • Traceable, permissioned operations
🔔

Open ecosystem

REST API plus Webhook subscriptions to connect order systems, CRM, ERP and your own apps so the AI can answer real business questions.

  • Taobao / JD / PDD / Douyin order flows
  • Session, ticket and CSAT events
  • Full developer docs and sandbox
ARCHITECTURE

Capability stack that works together

Five layers, integrated out of the box — deepen them step by step as your business grows.

Intake layerOmnichannel Ingress

Web widget, WeChat Official Account, Mini Program, WeCom, Douyin DMs and voice calls normalized into a single conversation bus.

WebWeChatMini ProgramWeComDouyinVoice
AI engine layerAI Engine

Knowledge retrieval plus LLM generation, with intent recognition and similar-question learning for answers that are both accurate and natural.

Knowledge baseLLMIntent detectionAuto learning
Orchestration layerFlow Orchestration

Visual canvas for multi-turn dialogs and business flows: conditions, API calls, approvals and scheduled triggers, all without code.

ConditionsAPI nodesApprovalScheduled
Agent workspaceAgent Desk

Omnichannel desk, smart queue routing, takeover, live translation, ticket flows with SLA and CSAT follow-up — all channels in one place.

Unified deskLive translationTicket SLACSAT
Insight layerBI & Integration

BI reports, hot-topic rankings, bot hit-rate analysis plus open APIs and Webhooks that feed service data back into decisions.

BI reportsTopic analysisOpen APIWebhook
COMPARISON

Traditional support vs. TianKey Cloud

DimensionTraditional / early botsTianKey Cloud
Response speedQueued waiting, manual repliesInstant AI answers, 24/7
ChannelsSingle phone line or web windowWeb / WeChat / WeCom / Douyin / voice in one desk
Answer qualityRigid keyword matchingKnowledge base + LLM, learns similar phrasings
Human handoverNo standard, customers lostRule-based, reason-tagged, auditable
Complex flowsQ&A only, no processVisual flows with API and approval nodes
PersonalizationOne script for everyoneMulti-bot routing with isolated knowledge
Cost structureHeadcount grows linearlyAI covers 80%+ of repetitive queries
Go-live timeWeeks of procurement and setupSaaS onboarding, one line of code for web
ScalingHire more people and devicesElastic SaaS, buy top-ups as needed
AnalyticsGut feeling, no metricsMulti-dimensional BI with scheduled reports
IntegrationData silo, no business linkageOpen API + Webhook for orders / CRM / ERP
SECURITY

Security & compliance, stated clearly

🧱

Tenant-level isolation

Company data stays invisible to other tenants — knowledge, transcripts and contacts are separated with clear permission boundaries.

🔒

Secure transfer & storage

HTTPS everywhere, encrypted sensitive configuration, full operation logging and traceable anomalies.

🚫

Sensitive words & compliance

Automatic sensitive-word interception plus compliant script libraries for finance, healthcare and other regulated industries.

👥

Granular permissions

Admin, agent and QA roles separated, with menu, data scope and action permissions assigned per role.

📜

Standard handover

Following GB/T 47746-2026: AI identity disclosed, human route always available, service process auditable.

🏠

On-premise option

For sensitive or internal-network requirements, the Standalone plan deploys on dedicated servers with fully private data.

FAQ

Common questions about our capabilities

What if the AI answers incorrectly?
Answers come from the knowledge base first and the LLM second. Unmatched questions are collected into an "unknown questions" pool; once an operator adds the official answer it is reused immediately. Handover to a human is always one click away, so accuracy converges week by week.
Do we have to change our systems to integrate?
No. A single line of JS brings the web widget live; WeChat, Douyin and WeCom only need credentials configured in the dashboard. To let the AI answer order or shipping questions, connect them via API nodes or Webhooks.
Will customers get stuck talking to a bot?
No. Handover rules cover negative sentiment, explicit human requests, repeated misses and exhausted quota, and both the reason and the process are logged in line with GB/T 47746-2026.
Can different business lines have different bots?
Yes. Multi-bot routing supports channel, keyword and customer-tag rules, each bot keeping its own knowledge base and persona.
What happens when conversation volume spikes?
Nothing stops. Quotas count per month; once used up, top-up balance automatically covers extra requests at ¥0.03 each and carries over to following months. Add credit ahead of campaigns when needed.

Experience the eight advantages yourself

¥9.9 trial (one year) · online activation · one line of code for the web widget

Start with ¥9.9 View Pricing