AI-native business
process management
The LiveBPM workflow platform runs your processes reliably and keeps a complete audit trail. The LiveAI agent foundation adds understanding, extraction and judgement. Since 2001 we have done one thing: manage business processes properly.
Meal expenses are 38% above this department's six-month average and include two sequentially numbered invoices. Review suggested. Basis: Travel & Entertainment Policy, clause 4.2.
Execution must be deterministic. Judgement can be intelligent.
A conventional system is capped by the number of rules you thought of in advance. An AI-native system is capped by how much business context it can understand. We combine the two rather than choosing between them.
Reliable execution
State changes, amounts, approval records and document routing are all handled by a deterministic process engine. The end-to-end audit log is where compliance and audit actually land.
Intelligent augmentation
Document extraction, clause comparison, cross-system querying and anomaly detection. Agents attach an evidence-backed recommendation to the process instance as structured data.
Human on the loop
The AI proposes; people keep the right to override. Every acceptance and every override is recorded and becomes your organisation's own judgement asset. Autonomy is earned from data, not assumed.
One guarantees execution. One supplies judgement.
The two are physically decoupled: separate processes, separate connection pools, separate call stacks. When the AI is unavailable the business path is still complete.
LiveBPM
A mature self-hosted BPM platform on .NET + Vue, with SaaS multi-tenancy, on-premise deployment and eight interface languages.
- BPMN 2.0 engine: countersign, add-signer, withdraw, transfer, delegate, overdue handling
- Low-code forms: drag-and-drop design, sub-tables and formulas, responsive across devices
- Single sign-on, organisation permissions, notification centre, reporting dashboards
- REST API covering process instances, to-do items, forms and approver routing
LiveAI
Eight layers of capability, from model access and data classification through to human-machine collaboration and governance, adding judgement at every step of a process.
- Multi-model routing: OpenAI-compatible protocol, sensitive data on a dedicated channel
- Extraction with confidence routing; anything uncertain is marked as such, never invented
- Hard rules implemented as code: no hallucination, reproducible, numbers never model-generated
- Four authorisation levels L0–L3; review, approval and audit all flow into the process engine
A deterministic skeleton with cognitive augmentation
Five layers. The intelligence layer and the business layer scale independently, because inference is compute-bound and business logic is I/O-bound.
A unified tool gateway instead of point-to-point integration
External systems are wrapped once as standard tools for agents and business modules to call. The number of interfaces drops from consumers × systems to systems. Authentication, auditing, rate limiting, caching, compensation and contract management are handled once, centrally.
Three levels of degradation, so AI failure never stops the business
A single failing agent hides only its own recommendation. If the model service is down the system falls back to pure rules. If an external data source is down the latest cache is used and clearly marked as stale. The business path stays complete in every case.
Start with one document, cover the core of the business
Expense claims
Invoice recognition and duplicate detection, budget commitment, automatic routing when limits are exceeded, with AI surfacing the applicable policy clause.
Procurement and contracts
Requisition, quotation, order, contract approval and payment tracking in one line, with clause comparison and risk flags.
Budget control
Preparation, commitment, adjustment and execution analysis, with available balance checked live at each step.
HR processes
Onboarding to offboarding, leave and overtime, performance reviews, with multi-organisation routing and delegation.
IT and admin services
Resource requests, asset issue, meetings and vehicles, with a standard service catalogue and SLA reminders.
Manufacturing systems
SRM, CPQ, QMS, MES and OMS built on the two platforms, sharing one engine and one agent foundation.
Ten AI scenarios, without a single new line in the foundation
A multinational commercial real estate services group: five departments, sixty frequent users, seven cities, fifty-three requirements. All eighteen agents were assembled by configuration.
Offline validation on anonymised synthetic samples; model outputs are frozen and the scripts are reproducible. A formal engagement requires a blind test on your own real samples.
Three principles that decide how we build
Protocol first, no vendor lock-in
Everything speaks the OpenAI-compatible protocol, so the model layer is replaceable. Changing model does not mean rewriting the application.
Additive only, non-invasive
We do not modify your systems of record; we add alongside them. Upgrades are reversible, so IT does not carry the risk of going live.
The foundation stays generic
Business logic lives in configuration. From a certain point on you extend the system yourself, without depending on us.
FAQ
A few common questions about the company, the products and the AI.
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.
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.
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.
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.