Junpu — powering the intelligent upgrade of enterprise operations

Junpu Teamify
A bespoke AI workspace for your enterprise

Rebuild your team around AI and automate the entire chain.

Core data stays on-premise, guarded by a data-language layer and a one-way valveData privacy
Automate repetitive work so your team gets 60% more time back for work that creates new valueTeam efficiency
Our in-house Agent Harness executes strictly within business constraints — no AI hallucinationA partner you can trust
The Factify data hub breaks down silos and lifts cross-department collaborationData as an asset
junpu.org / AI team workspace
LIVE
CEO · AI
Decision hub · task dispatch
🎧
Service AI
Customer service, pre-sales
Online
📦
Procurement AI
Supply chain, sourcing
Working
💼
Finance AI
Financial analysis, reconciliation
Online
📈
Sales AI
Market development, account care
Done
Live task flow8 in progress
08:00Service AIProcessing 12 emails
08:02Procurement AIComparing quotes from 3 suppliers
08:05Finance AIGenerating the monthly cash plan

AI employee org chart · a CEO decision hub orchestrating specialised AI employees in parallel

Figures as of September 2026
¥10B+
Peak GMV served
50+
Enterprise clients
180+
Agents and workflows
Customer service · LIVE DEMO

Let AI take over the support inbox — not a single email missed

Every day before 8:00, AI sorts the inbox, classifies each message, drafts replies and scores its confidence.
Engineers only review. One support email, end to end, in 23 seconds.

8:00
AM · auto kick-off
6types
Auto-classified
87%
High-confidence drafts
junpu.org / Support workspace
LIVE
📥Inbox triage17:30 → 08:00
✕Flash Sale CentreBuy now! Last 3 hoursAd
✕Security AlertUnusual activity on your account — verify nowPhishing
✓Mr. ZhangRe: Order #20260726 cancellation requestValid
✓Ms. LiRefund & return — item arrived damaged (3 emails sent)Valid
✓Hans MüllerS9 installation issueValid
12 spam emails filtered out · 8 valid
Procurement · LIVE DEMO

The procurement chain
6 steps, fully automated

From raising a request to reconciling invoices — six stages, each AI-augmented. Watch the traditional process run side by side with the AI one, and see how days of work compress into hours.

6
Process steps
5
Departments
80%
Efficiency gain
junpu.org / Procurement workspace
LIVE
Step 1 / 6

Request intake

✕Traditional process
  • Fill in the purchase request by hand (model, quantity, target price)
  • Back-and-forth emails to confirm missing details, 1-2 days lost
✦AI-augmented process
  • ✓A standardised form auto-fills and validates, prompting for any missing field
  • ✓Pulls historical specs, deal prices and lead times for the same model as reference
  • ✓The request is filed in one click and a unique ID syncs to the procurement workspace
Efficiency gain1-2 days → 10 minutes

Customer stories

From user growth to business operations

We serve manufacturing, supply chain and e-commerce companies — with customer stories spanning user growth through to business operations.

Scenarios · Customer interviews · Milestone results
01ANONYMIZED BUSINESS CASE

Food brand · Omnichannel commerce

A food e-commerce brand

Operating data stops at the report — six AI assistants work on it together

Across new-product, sales, production, media-buying and finance scenarios, we unified the data definitions scattered across channels so AI could move from insight to executable operating coordination.

Multi-channel data intake
Operating semantic layer
6 AI assistants
Confirmed project scopeFrom product development to channel reconciliation
  • Data ETL and metric governance
  • New product, sales and production planning
  • Ad analysis and channel reconciliation
Real business scenario · anonymisedDiscuss a similar scenario
02ANONYMIZED BUSINESS CASE

Enterprise services · Supply chain

A supply-chain services company

From supplier sourcing to financial settlement — an executable AI supply chain

Connecting procurement, commercial, bidding and finance workflows, we configured specialist AI employees on top of enterprise data and business rules to support cross-department coordination.

Business systems and documents
Supply-chain data hub
4 types of AI employee
Confirmed project scopeCovers the core supply-chain workflow
  • Supplier sourcing and smart price comparison
  • Order, logistics and bidding coordination
  • Invoice verification and financial reconciliation
Real business scenario · anonymisedDiscuss a similar scenario
03ANONYMIZED BUSINESS CASE

Cross-border trade · Platform commerce

A cross-border e-commerce group

Handing reconciliation, performance and analysis in cross-border operations to AI

For multi-platform, multi-role and complex financial definitions, we turned order, organisational and operating data into traceable analysis and execution flows, cutting repetitive data work.

Order and operating data
Unified metric definitions
Analysis and execution assistants
Confirmed project scopeFocused on high-frequency operating data work
  • Automated reconciliation and variance tracking
  • Organisational performance assessment
  • Financial analysis and data assistant
Real business scenario · anonymisedDiscuss a similar scenario

Customer acquisition stories

Products differ —
finding the right buyer is the shared starting point

Three industries, three routes to new customers.
Concrete progress, drawn from customer interviews and business data.

A Foshan environmental technology company customer interview, including first-month results
Customer interview · first-month review
Commercial refrigeration01

A Foshan environmental technology company

Matching commercial refrigeration equipment to the right overseas channels

Slush and ice-cream machines serve restaurants, hotels and many other commercial settings, so buyers are scattered and manual filtering is slow. The company ran prospecting around product categories, focusing on the distributors and manufacturers that fit, to open up overseas partnerships.

Target buyers

Distributors and manufacturers of slush and ice-cream machines

Milestone results 7 months in

Target accounts
1,107
High-intent inquiries
67

Focused on overseas channel development for commercial equipment

A Jiangsu electronic materials company customer interview, including first-month results
Customer interview · first-month review
Industrial specialty tape02

A Jiangsu electronic materials company

Focusing on specialist channels gives industrial tape a clearer direction

Industrial tape has broad applications, but scattered end-user demand is not always suited to bulk supply. The company focused prospecting on specialist importers, distributors and wholesalers, looking for channel buyers that fit, then pushed inquiries through sampling to closing.

Target buyers

Specialist tape importers, distributors and wholesalers

Milestone results 4 months in

Target accounts
1,102
High-intent inquiries
43

One order closed

A Hebei machinery manufacturer customer interview
Customer interview · machinery
Custom machined parts03

A Hebei machinery manufacturer

Reaching overseas buyers for machined parts, segment by segment

Non-standard parts have long procurement cycles, and searching for buyers and writing outreach one by one is slow and hard to match to a niche. Through proactive outreach aimed at machinery and parts makers, the company opened up overseas inquiry opportunities for its custom business.

Target buyers

Makers of machinery and agricultural machinery parts

Milestone results 10 months in

Target accounts
5,659
High-intent inquiries
53

Some buyers have reached the factory-audit stage

About these resultsEach case ran for a different period. "High-intent inquiries" follows the definition used in the case material and does not equal closed orders.

Discuss my acquisition scenario

End-to-end AI acquisition

AI that helps exporters win customers

From spotting the market to closing the deal — let global customers find you.

Connecting global opportunitiesAccelerating growth

Market insight

Spot high-potential markets and buyers

  • Global demand analysis
  • Industry trends and opportunity spotting
  • Target country and region screening
  • Ideal customer profile generation

Precision prospecting

Mine real buyers across channels

  • AI search and harvesting
  • Company data deduplication and cleansing
  • Contact and decision-maker identification
  • A continuously updated customer database

Lead scoring

Filter for the most valuable opportunities

  • AI scoring and intent detection
  • Company strength and credit checks
  • Purchase needs and historical behaviour analysis
  • A prioritised follow-up list

Smart outreach

Personalised, multilingual contact

  • AI-written personalised outreach emails
  • Automatic multilingual translation and polishing
  • Email · LinkedIn · WhatsApp multi-channel reach
  • Automated follow-ups and reply reminders

Pipeline management

From lead to order, followed through

  • A visual sales funnel
  • Interaction history and reminders
  • Email · samples · contract management
  • Team collaboration and performance tracking

Sustainable growth

Data-driven repeat and referral business

  • Customer lifecycle management
  • Repeat purchase and referral mining
  • Market and customer data accumulated
  • Continuously refined acquisition strategy
1M+
Companies in the global database
3x
Efficiency in winning quality opportunities
50%
Sales follow-up time saved
Sustained growth
Data-driven performance
Want AI to take over your acquisition pipeline?See the export acquisition delivery plan

Teamify AI · Core capabilities

The core capabilities of an enterprise AI team

Every capability is battle-tested, so your agents become partners.

Factify data hub

The foundation AI teams work on

Data hub

Your enterprise data centre — business data, industry data, API data and API interfaces.

Data semantic layer

Your enterprise gains a digital brain with common sense that understands the business.

Vast business data · Data as an asset · A layer of facts

Business capability centre

The AI business capabilities your company needs

Business SOP consulting

We abstract the business capabilities an AI team needs from your requirements and build them into a capability centre.

Lightweight delivery

No complex, costly graph-database cluster — Factify deposits business entities while Agent Skills execute business processes.

Deep vertical focus · Bespoke delivery

Teamify collaboration system

Let the AI team work on its own

Multi-agent collaboration

Run many AI agents at once — coding, analysis, operations — orchestrated from a single console.

Business automation

Your AI employees run tasks on schedule and request approval for anything important.

Parallel execution · Automated operations

Knowlify knowledge hub

Business knowledge, systematised and self-evolving

Systematised knowledge

Weaves fragmented knowledge into a dynamically updated business mesh that supports logic-driven inference.

Native terminal

Edit and deposit knowledge straight from the terminal, or feed it to agents so they keep evolving.

Fixes information overload and thin logic · Grows and evolves on its own

Architecture overview

Four layers that put AI to work in the business

This is not just another chat window. From enterprise data and the work platform to specialist skills and the AI team, every layer serves the business outcome.

AI capability architecture
Business outcomeData foundation
Top layer04

BUSINESS ORCHESTRATION

Enterprise AI team

Builds digital-employee teams around business goals, with automatic task breakdown, collaborative execution and delivery.

Goal breakdownCross-role collaborationResult delivery
Organisation & operations
Layer 303

READY-TO-USE SKILLS

Ready-to-use AI skills

Customer service, marketing, sales, HR, finance, legal and other mature capabilities — configure and use, and keep evolving.

Industry templatesSpecialist skillsContinuous evolution
Scenarios & capability
Layer 202

AGENT WORK PLATFORM

AI work platform

Create, orchestrate, debug and publish digital employees so complex AI applications iterate fast.

Create agentsVisual orchestrationDebug and publish
Build & run
Foundation01

DATA & SECURITY FOUNDATION

Enterprise data and security foundation

Connects internal systems and enterprise knowledge with data governance, permission isolation and end-to-end auditing.

Data intakeKnowledge governancePermission auditing
Data & governance

Why the architecture matters

Turns AI from a peripheral tool into organisational capability

Business goalExecutable task
Single assistantCollaborating team
Generic answerEnterprise delivery

Security × evolution

Secure by design, always evolving

An enterprise-grade security foundation protects your data while AI employees keep improving on the job.

Security foundation

Encryption

AES-256 across transport and storage, with keys in your hands

Permission isolation

Three-tier control by role, department and project, on least-privilege principles

Private deployment

Deploy on-premise — data never leaves your domain or goes out

Evolution engine

Self-learning

Extracts knowledge from every task and keeps refining its strategy

Skill evolution

Install a new skill pack and gain the capability in seconds, no retraining

Collaborative boost

Multi-agent experience compounds, lifting the whole team's efficiency

Permissions & team management

The right people, with the right access, doing the right work

An enterprise permission model × flexible team collaboration — from org structure to resource access, every layer is controllable and auditable.

Permission model

Role permissions (RBAC)

Three-tier control by role, department and project on least-privilege principles, with dynamic authorisation and permission inheritance.

Data permissions

Row-level and field-level isolation with sensitive fields masked, so members only see what they should.

Resource permissions

Access and operation rights for agents, workflows and knowledge bases are configured separately, for fine-grained control of every asset.

Team collaboration

Organisation structure

Multi-company, multi-department and cross-team hierarchies, supporting matrix organisations and virtual teams.

Member management

Invitations, role inheritance and bulk configuration, with permissions adjusting automatically when members change.

Collaboration spaces

Shared sessions, team agent pools and cross-team collaboration, with limited access for external suppliers and customers.

Role permission matrix

Default permissions of each role over each resource, adjustable as needed

RoleAgentWorkflowDataAudit log
Administrator
Full control
Manage
Manage
Manage
Manage
Department lead
Department scope
Manage
Manage
View
View
Member
Day-to-day use
Operate
Operate
View
None
Guest
External collaboration
View
None
None
None
3
Permission tiers
4
Role types
100%
Action auditing

Built in-house: three AI suites for enterprise intelligent operations

From data to knowledge, from collaboration to execution — Junpu's systems cover every core step of putting AI to work in an enterprise

Teamify

Human-AI collaboration system

Brings AI employees into your teams so they work efficiently alongside people, with task assignment, execution oversight and continuous improvement.

  • AI employee scheduling and task orchestration
  • Human-AI workflow engine
  • Execution audit and performance tracking

Factify

Data hub · Data-semantics AI suite

Breaks down data silos and builds a unified data hub, with a semantic layer so AI genuinely understands enterprise data.

  • Automatic enterprise data-lake construction
  • Semantic understanding and data access control
  • Connects ERP/CRM/OA and other sources

Knowlify

Enterprise knowledge hub

Deposits business knowledge into a searchable, reason-capable system so AI carries real industry expertise.

  • Automatic extraction and deposit of business knowledge
  • Knowledge graph and semantic search
  • Continuous learning and knowledge refresh

Defence in depth

Three layers of security, defence in depth

From deployment to runtime to monitoring, every layer is fully protected.

1
Layer 1

Private deployment

Runs on your own servers, private cloud or hybrid cloud. Data stays entirely in your hands and never passes through a third party.

On-premise / private / hybrid cloud
Data never leaves your domain
100% data sovereignty
2
Layer 2

Isolated execution

Each agent runs in its own sandbox with isolated memory and filesystem. Zero interference and zero leakage between tasks.

Isolated sandbox
Memory isolation
Filesystem isolation
3
Layer 3

Defence system

Behaviour is monitored in real time and privilege escalation blocked automatically, with full audit logs for compliance.

Real-time behaviour monitoring
Automatic privilege blocking
Complete audit logs
User request
Encrypted transport
Sandboxed execution
Result returned

Evolution path

Four steps from tool to organisation

AI employees are not static tools. They grow in practice, strengthen through collaboration, and end up an inseparable part of your organisation.

01
01

Advisor-led discovery

Our FDE advisors define AI employee roles from your processes, configure them and set them to work — no programming knowledge required.

02
02

Skill enablement

Install skill packs for your AI employees: data analysis, content creation, customer service — capability expands with the business.

03
03

Team collaboration

Several AI employees form a team and divide work automatically. A sales agent hands a qualified lead straight to a service agent.

04
04

Continuous evolution

AI employees learn from every task and keep getting more efficient. Your team gets stronger and smarter the more you use it.

Three marketplaces, endless possibilities

Tools, models, skills — three marketplaces cover everything you need

Tool marketplace

Hundreds of automation tools ready to install in one click, covering data analysis, content generation, customer management and more.

Model marketplace

Switch freely between GPT-4, Claude, Qwen and other leading models. Match the best model to each task and balance quality against cost.

Skill marketplace

Community-contributed skill packs give your AI employees new capabilities in seconds, with custom creation and sharing supported.

The Junpu deployment path

Live in three steps

From first conversation to launch — your first AI employee can start working in as little as one week.

3 hours

Discovery

A 3-hour session with our AI advisors to understand your business scenarios and needs.

Pain-point analysis
Mapping AI use cases
ROI estimate
3-5 days

Custom configuration

We tailor AI employee roles, skills and workflows to your processes — with hands-on help throughout.

Role definition and configuration
Skill pack installation and tuning
Workflow orchestration and testing
Same day

Go live

Your AI employees start on real tasks immediately, backed by round-the-clock technical support.

Canary release and validation
Performance monitoring and tuning
Ongoing technical support

The Junpu ecosystem

Integrates seamlessly with your stack

Your AI employees already know the tools you use. Plug and play, zero migration cost.

Cloud providers
Alibaba Cloud
Tencent Cloud
Huawei Cloud
Baidu AI Cloud
AWS
Azure
Google Cloud
Volcano Engine
Alibaba Cloud
Tencent Cloud
Huawei Cloud
Baidu AI Cloud
AWS
Azure
Google Cloud
Volcano Engine
Domestic chips
Huawei Ascend
Cambricon
Hygon
Biren
Moore Threads
Iluvatar CoreX
Enflame
MetaX
Huawei Ascend
Cambricon
Hygon
Biren
Moore Threads
Iluvatar CoreX
Enflame
MetaX
Hardware
NVIDIA
AMD
Intel
Qualcomm
ARM
Apple Silicon
Google TPU
Groq
NVIDIA
AMD
Intel
Qualcomm
ARM
Apple Silicon
Google TPU
Groq
Collaboration
WeCom
DingTalk
Feishu
Slack
Notion
Microsoft Teams
Lark
Zoom
WeCom
DingTalk
Feishu
Slack
Notion
Microsoft Teams
Lark
Zoom
200+
Ecosystem partners
30+
Chip adapters
15+
Cloud providers
100%
Domestic stack compatible

Security & compliance

Enterprise-grade security, certified compliance

A stack of international and domestic security certifications standing behind your data.

ISO 27001

Information security management certification, keeping data governance aligned with international standards.

ISO 9001

Quality management certification, keeping product and service quality consistently on target.

SOC 2

Service organisation control report validating security, availability and confidentiality controls.

MLPS Level 3

Level 3 of China's Multi-Level Protection Scheme for information systems, meeting critical-infrastructure security requirements.

Proprietary IP

Core algorithms and system architecture developed entirely in-house, backed by multiple invention patents.

Book a consultation on your AI upgrade

A 30-minute business consultation to map out your AI rollout

How it works

  • 1

    Tell us what you need

    Share your business context and what you want to solve

  • 2

    Add us on WeChat

    Scan the QR code to add our customer-service account

  • 3

    Pick a time

    Confirm a slot for a 30-minute consultation with a specialist

After submitting you will receive instructions to add us on WeChat

Frequently asked questions

Common questions about building an AI digital-employee team

Still have questions?

Talk to an advisor