Agent management
Design, deploy, monitor and govern enterprise AI agents from one place — including role, authority limits, escalation triggers and lifecycle state.
Enterprise AI Platform
LyndonAI — the enterprise AI and ontology platform.
LyndonAI is Cyberwisdom Group's enterprise-level AI platform: the place where agents, enterprise ontology, context and governance are managed together rather than assembled from parts each deployment.
What is LyndonAI?
LyndonAI is Cyberwisdom Group's <b>enterprise-level AI platform</b>, and the enterprise AI agent and ontology management platform at the centre of the group's portfolio. It is listed by the Hong Kong government Smart Government Innovation LAB as an enterprise AI solution. Within Cyberwisdom's wizBank 8.0 platform generation, LyndonAI operates alongside the UpLearn learning management system, MetaSim business and sales simulation, and Kora knowledge management as four connected systems — supported by VibeChat, an intelligent AI knowledge copilot, and Optima, an enterprise human workflow and agent system. LyndonAI's role is to give enterprises a single governed place to manage AI agents, the business ontology they reason over, and the context they are given.
The platform
One governed place for agents, ontology and context.
The problem with assembling enterprise AI from parts is not the parts. It is that authority, context and evaluation end up in three places, and nobody can answer what the system was actually given.
Ontology management
The business ontology agents reason over, managed as a first-class asset with ownership, versioning and change control rather than as configuration buried in a repository.
Context assembly
Governed retrieval and knowledge grounding, with authority and freshness tracked per source, so what the model is given can be explained.
Workflow composition
Human steps and agent steps composed into executable flows — handled natively through Optima rather than bolted onto a chatbot.
Evaluation & observability
Reliability, cost and latency tracked per run, with the ability to reconstruct any decision the system made.
Governance & evidence
Risk classification, control enforcement and evidence capture integrated into operation, aligned to ISO 42001 through ComplyAI.
The connected systems
LyndonAI within the wizBank platform family.
The platform generation is deliberately a system of connected parts rather than a single monolithic tool.
wizBank 8.0
The flagship Human Machine (AI+Data) Capital platform. Four core systems operate together, with a Capability Ontology and capability graph connecting what the organisation knows to what it can do.
UpLearn LMS
The enterprise learning management system — course and assessment, positioned as an agent-ready LMS. It is also where competence, certification and compliance evidence live.
MetaSim
Enterprise business and sales simulation; a sales and soft-skills AI simulation providing an enterprise behaviour digital twin for rehearsal and assessment.
Kora KM
Ontology-based knowledge management — the institutional knowledge layer that both the learning estate and the AI estate are built on.
Deployment
How it is deployed.
| Model | What it suits | How it runs |
|---|---|---|
| Cloud (SaaS) | Standard enterprise deployments with multi-tenant efficiency | Cyberwisdom-hosted, managed by the group |
| Dedicated cloud | Organisations requiring data isolation with managed operations | Single-tenant environment, operated under agreed service levels |
| On-premise | Regulated or sovereign environments requiring full control | Run inside the organisation's own infrastructure, under its own controls |
| Hybrid | Organisations running across regions or with data-residency constraints | Workload placed according to data sensitivity and latency requirements |
Why it exists
What the platform is actually for.
The point is not the platform. The point is the operating model.
Technology is the least interesting part of an enterprise AI transformation. The platform matters because it makes the operating model enforceable rather than aspirational.
- <b>It makes authority explicit.</b> When an agent's permissions, context sources and escalation rules live in one managed place, they can be reviewed, changed and evidenced.
- <b>It makes reuse real.</b> The ontology, context layer and evaluation harness built for one use case are inherited by the next — which is where the economics change.
- <b>It makes governance operable.</b> Controls enforced at runtime beat controls described in a policy, because they cannot be quietly bypassed.
- <b>It makes improvement measurable.</b> With evaluation and observability native, reliability becomes a tracked number instead of an anecdote.