Architecture & Security — Evaluate a HEPO Deployment
A technical evaluation guide to HEPO's operating records, execution and evaluation responsibilities, integration, local inference, access, and deployment assurance.
HEPO is enterprise software for people, culture, and business processes, with Simulation & Gamification as its third offering. This reference is for architects, security reviewers, infrastructure teams, and implementation partners evaluating how those offerings fit an organisation's systems and operating requirements.
The evaluation should follow a representative piece of work: its records, execution, interpretation, decisions, and resulting changes. That makes the architectural responsibilities and deployment boundaries concrete.
The application, engine, and operating records
The application presents work through shared interface patterns and registered surfaces. Nebulus provides the server-side orchestration and worker environment. The Operational Core holds the connected people, work, events, communication, knowledge, and other records used across the offerings.
Process execution and trait evaluation have different jobs. Execution coordinates defined operations, dependencies, results, and human gates. Evaluation interprets relevant evidence against definitions and produces observations or composite readings. The assistant provides conversational access to supported retrieval, navigation, and operations. A common platform lets them cooperate without making them interchangeable.
Typed contracts and generated application/server definitions reduce repeated interface work. Deployed versions and custom extensions still require compatibility checks. The System Layer maps the services behind those contracts and operations.
Follow the P&C sequence without collapsing its decisions
| Part | Evaluation responsibility |
|---|---|
| Catalogue | Establish the organisation-owned definitions and observation scope |
| Engine | Evaluate relevant evidence against those definitions |
| Indices | Examine composite readings, trends, and available uncertainty context |
| Recommendations | Inspect a proposed response and its basis |
| Programmes | Undertake the development work people decide to pursue |
| Flourishing and wellbeing review | Consider subsequent evidence, activity, and participant experience |
This sequence connects evidence to action and review. It does not mean every stage is automatically computed from the previous one. People choose responses and undertake work; a changed reading alone does not establish that an intervention caused it. A recommendation is a proposal informed by evidence, not the evidence itself or an authorised action.
Establish source ownership and exchange direction
The ERP or HRIS can remain authoritative for its existing domain. HEPO holds its own operational records and receives agreed information through configured connectors. Where a workflow needs outbound delivery, it requires a supported writable adapter and a separately authorised path.
Use the Data Integration reference for connector, mapping, schedule, and delivery responsibilities. Use the Process example to follow how the exchange participates in a decision. The implementation must define identifiers, ownership, reconciliation, failure handling, and what counts as confirmed delivery.
Inspect security and data boundaries
The Security Model explains shared identity, roles, permissions, enforcement, and audit review. Client presentation and server protection are distinct responsibilities. Test the relevant operation under allowed and denied contexts, including record and field restrictions where the use requires them.
Data Integrity distinguishes business records, definitions, audit, analytics, and platform data. In the server architecture, database schemas separate concerns; a schema boundary is logical separation, not necessarily separate physical infrastructure. A shared multi-tenant arrangement requires its own verified routing and isolation across data, files, reporting, and background execution.
Audit inspection should establish actual coverage, actor and process context, access, retention, and protection of the stored account. It should also distinguish a person's decision from scheduled execution or an external credential used to carry out the work.
Review deployment and local inference
The deployment places services, data, and local inference within the organisation's agreed infrastructure boundary. Object storage holds documents and media under the selected operating arrangement. The Deployment reference owns topology, operation, updates, backup, recovery, and handover.
Local AI describes inference within that boundary. Review the configured endpoints, model capacity, permitted context, and network path with representative requests. Local processing does not remove the need to evaluate the meaning and quality of a model's output.
Encryption settings, key management, retention, support access, recovery evidence, and any required audit or certification reports must be assessed for the actual deployment and engagement. Product architecture alone does not establish certification or satisfy an organisation's complete compliance obligations.
Bring the evaluation to a working result
Ask the implementation team to follow one process from source record to result, including a human decision, an invalid input, a denied action, and an unavailable dependency. Inspect the relevant trace and any supported outbound delivery. For a P&C application, follow a reading to its definition and evidence, then distinguish the proposed response from the decision to act.
The organisation supplies its requirements and responsible reviewers. Implementers configure and demonstrate the environment. The delivery partner coordinates the engagement, acceptance evidence, and handover.