Company
What kind of company is DLTECH?
Shanghai Donglan Information Technology Co., Ltd. (DLTECH) is a company registered in Shanghai. We are a digitalisation and AI partner for enterprises and public institutions. Built on our own LiveBPM process platform and LiveAI agent foundation, we cover the whole path from putting processes online, through complete business systems, to industry platforms and applied AI. We build the LiveBPM workflow platform and the LiveAI agent foundation.
How are LiveFlow and LiveBPM related?
LiveFlow was our earlier workflow product. LiveBPM is the current platform: the same engine experience rewritten on .NET + Vue, with SaaS multi-tenancy and on-premise deployment.
What does the team look like?
We run a platform engineering team responsible for product development, delivery and support, plus a dedicated agent team that validates and ships AI scenarios.
Are LiveBPM, workflowww registered trademarks?
Yes. LiveBPM, workflowww are registered trademarks of Shanghai Donglan Information Technology Co., Ltd.. They may not be used in goods, services, domain names or marketing materials without written authorisation.
LiveBPM workflow platform
What is LiveBPM?
LiveBPM is our enterprise workflow / BPM platform. It provides a BPMN 2.0 process engine, a low-code form designer, organisation and permission management, rule-based routing, a notification centre, reporting dashboards and a REST API. It is responsible for reliable execution and a complete audit trail.
Which process operations does LiveBPM support?
Start, assign, withdraw, direct-send, hold, pause, stop, resume, countersign, add-signer, notify, transfer, delegate, archive, out-of-office delegation, substitute handlers, shared tasks, overdue reminders and automatic overdue handling, scheduled start. All of these are built into the engine, not scripted per project.
What can LiveBPM integrate with?
Single sign-on against LDAP / Active Directory and mainstream OAuth2 / OIDC identity providers; ERP, HR, CRM and finance systems through a unified tool gateway and adapters; a REST API covering authentication, process instances, to-do items, form data and approver routing; webhook events; and messaging through WeCom, DingTalk and Feishu.
Can LiveBPM work with AI office assistants?
Yes. Starting a process, querying work items, reading and writing form data and resolving approvers are wrapped as standard tools and registered through each assistant's plugin or skill interface, so people can check progress, pick up tasks, file requests and approve from a conversation in WorkBuddy, 豆包 Work and 千问办公. The assistant is only the entry point: permissions follow single sign-on, and state changes and approval records are still written by the process engine.
Does LiveBPM support multiple languages and mobile?
Eight interface languages, selectable per user. The web interface is responsive, and the mobile view embeds into the WeCom, DingTalk and Feishu workplaces.
What is the technology stack and how is it deployed?
.NET on the back end, Vue on the front end, decoupled, delivered as containers. Available on-premise (your data centre or private cloud), as SaaS multi-tenant, or hybrid with the process in the cloud and sensitive data and models kept local.
How much can be built without writing code?
Process modelling, form design and report generation are all graphical. More than 90% of processes, forms and reports are built without code; the remainder is handled through open APIs and source-level extension.
LiveAI agent foundation
What is LiveAI?
LiveAI is our enterprise agent foundation. It has eight layers: model access and routing, data ingestion and classification, anti-hallucination grounding, information extraction, a rule engine, tool wrapping and agents, human-machine collaboration, and governance and observability. It adds extraction, review, scoring and question-answering to each step of a process.
Can the AI change business data or approve on its own?
No. LiveAI never changes business state. It can read process context, write a recommendation with its evidence, and create a work item for a person to confirm. In L0 authorisation it may trigger reminder-type helper flows. State changes, amounts and approval records are always executed by the LiveBPM engine under human authorisation.
What happens when the AI is wrong or hallucinates?
Every judgement carries an evidence chain and its data sources. Numeric conclusions are computed deterministically rather than generated by a model. Critical judgements can be cross-checked by two models. Anything at L2 or above requires explicit human confirmation. In our ten-scenario validation, confidence-based routing and the rule engine together intercepted 75% of the error cases before they reached a user.
Does the business stop if the AI service is unavailable?
No. The intelligence layer and the business layer are physically decoupled, with three levels of graceful degradation: a single failing agent simply hides its recommendation; if the model service is down the system falls back to pure rules; if an external data source is down the most recent cache is used and clearly marked as stale. Process integrity is guaranteed by the engine, not by the AI.
Does our data leave the company or train public models?
Fully on-premise deployment is supported, in which case no data leaves your environment. In API mode traffic passes through a model gateway that enforces masking, under a data processing agreement. The model layer speaks the OpenAI-compatible protocol, so sensitive data can be routed to a dedicated instance. Customer data is not used to train public models.
What is the four-level authorisation model?
AI actions are graded by consequence and confidence: L0 automatic execution, L1 silent recommendation, L2 explicit confirmation, L3 human-led. Most agents sit at L1 or L2. The level is widened or tightened based on measured acceptance rate, override rate and post-hoc accuracy, not on opinion.
Delivery and commercial
How do I request a demo, and how quickly do you reply?
Use the form on the contact page, email sales@livebpm.com.cn, or call 021-54890228. We reply within one working day. The demo is built around one of your own real processes.
How long does it take to validate an AI scenario?
A single-scenario proof of concept usually takes four to six weeks: offline validation on anonymised samples first, then a blind test on real samples, then agreement on the authorisation level for each judgement.
Which processes should we start with?
Start with processes that are mandatory, frequent and tied to money: expense claims, purchase requests, contract approval. They are used daily and affect people personally, which is what makes the change stick. Expand in stages from there.
Do you have a full suite of business systems as well as the platform?
Yes. Built on LiveBPM and LiveAI we offer SRM (supplier relationship management), CPQ (configure, price, quote), QMS (quality management), MES (manufacturing execution) and OMS (order management). They share one process engine, one organisation and permission model and one agent foundation, so they can be combined freely or adopted one at a time.
Do you support domestic (Xinchuang) software and hardware stacks?
Compatibility with domestic operating systems, databases and middleware is verified per project and documented in a compatibility report. Please confirm the specific list with us.
Glossary
- BPM (Business Process Management)
- A discipline and a class of software for modelling, executing, monitoring and continuously improving the business processes of an organisation.
- Workflow
- The executable form of a business process: tasks are routed to the right person or system by rule, and every step is recorded.
- BPMN 2.0
- The OMG modelling standard that describes tasks, gateways, events and swimlanes with one set of symbols. It is the modelling language of mainstream process engines.
- Low-code form
- A business form built by dragging and configuring rather than programming, with sub-tables, formulas, validation and responsive layout.
- AI-native BPM
- Treating AI as a cognitive layer across the whole process rather than one item on a feature menu: the engine handles deterministic execution while agents supply evidence-backed judgement at each step.
- Agent
- A unit of AI software that can read business context, call tools, carry out multi-step tasks and return a conclusion with a recommendation.
- Human-on-the-loop
- The AI analyses and proposes; a person keeps oversight and the right to override, while the cost of that oversight is reduced because the AI has already filtered and evidenced the case. Distinct from human-in-the-loop (a person acts at every step) and human-out-of-the-loop (fully automatic).
- Four-level authorisation
- L0 automatic execution, L1 silent recommendation, L2 explicit confirmation, L3 human-led, graded by consequence and confidence.
- RAG (retrieval-augmented generation)
- Retrieve relevant material from the enterprise knowledge base first, then let the model answer from what was retrieved, so conclusions trace back to source text and fabrication is reduced.
- Hallucination
- Model output that reads plausibly but is not true. Controlled with evidence chains, confidence-based routing, deterministic computation of numbers, and human confirmation.
- Confidence routing
- Every AI output carries a confidence score; anything below the threshold goes to human review instead of straight to the user.
- Audit trail
- A complete record of who handled each step, when, with what comment and against what data snapshot. Replayable and auditable.
- SaaS multi-tenancy
- One deployment serves many customer tenants, with logical data isolation and per-tenant provisioning.
- On-premise deployment
- The system runs in the customer's own data centre or private cloud; data does not leave the organisation.
- SSO (single sign-on)
- One authentication grants access to multiple systems, typically through LDAP / Active Directory or OAuth2 / OIDC.
- WfMC
- The Workflow Management Coalition, which publishes the workflow reference model and interface standards.
Last updated . When citing, please credit Shanghai Donglan Information Technology Co., Ltd., https://www.livebpm.com.cn/en/faq.html.
Want to see it running on your own process?
Book a 45-minute online demo. We build a working sample from one of your real processes and show how deterministic execution and AI recommendations work together.