Appendix 5.2 · Learning and Project Funnel

Oristella Learning and Project Funnel

Technical assignment | University game, qualifications and client demonstrations

Build a single learning platform with two connected entry routes. A student enters through a free, short game on a synthetic gold-company case. A company enters through a free, bounded demonstration on synthetic data and then a paid, scoped assignment. Both routes feed a supervised catalogue of tasks, regional knowledge and qualified people. The initial release must demonstrate the connection without claiming that training outcomes or sales are already proven.

1 Users and journeys

Student: choose a field, complete a short diagnostic, play a guided case, receive a personal learning path, enrol in a course, complete practical tasks, sit a supervised exam and receive a class-specific credential. Strong candidates may opt in to a talent pool, scholarship review, paid project work or a job interview.

Company: select a sample task, enter only fictional or public training inputs, see a non-downloadable draft balance, geological model, market analysis or document package, then request a defined paid pilot. A qualified specialist reviews every real deliverable. Demonstration data never silently become client data.

2 Common case and specialist tracks

The game asks the learner to build a management model for a junior gold company. Its four first stations are geology and drilling, finance and reporting, market and strategy, and licence documentation. Later tracks may cover mapping and geospatial data, mining planning, sampling and laboratory control, operations and logistics, legal and contract work, information engineering, AI-agent supervision, and education or client implementation. The track catalogue is modular; the first release should select only disciplines needed for the Guinea pilot.

Every track has an entry task, guided practice, a versioned rubric, an independent final task and a specialist reviewer. Real regional assignments enter the course only after permission, anonymisation and task approval. A thesis may solve an approved regional problem; its accepted result becomes a candidate knowledge object, not an automatic production rule.

3 Qualification ladder and authority

Class is awarded per discipline. A person may hold Class III in geology and Class I in marketing. Overall leadership requires a defined combination of credentials plus observed supervision, not a single token or test score. AI agents receive separately tested capability passports and no human qualification title.

4 Assessment, rewards and fairness

Score accuracy, source quality, explanation, correction of errors, responsible AI use and reproducibility before speed. Keep model-generated hints separate from unaided exam work. Sample and audit scoring by language and university to detect unfair exclusion. A human appeals panel reviews consequential decisions.

A proposed internal fee model from the mobile design dialogue uses illustrative deposits of $10 for students and $100 for other specialists, returned on successful completion, with optional paid tutoring and awards for exceptional work. These amounts remain service-design assumptions pending local legal, affordability and budget review. Public marketing must not promise a scholarship, award or job to every graduate. Entry to the talent pool requires consent and clear retention rules.

Scholarships, thesis awards, language bonuses and paid assignments depend on a published rubric, budget and available work. Payment for a contributed knowledge object requires a written rights and remuneration agreement; course participation alone does not transfer the learner’s intellectual property.

5 Knowledge and agent feedback loop

Each accepted practical result carries a source passport, reviewer, quality state, rights, version and local scope. Students can contribute new local information, including official requirements, field observations and documented reasons for rejected submissions. The central team compares such evidence across regions and updates a model only after validation and release approval.

For an agent, record task success, failures, overrides, citations and the human sign-off. A failed task creates a negative-knowledge case with the circumstances and counterexample. It does not self-modify a production agent. Accepted changes move through a test suite at the Guinea centre, then a controlled release to other regional centres.

6 Prototype screens and data

The HTML prototype needs a role choice; four playable sample stations; a visible input-to-output transformation; diagnosis and class pathway; course outline; sample exam; consented candidate portfolio; a company demo; and a request for a paid pilot. Use synthetic data only. The prototype must work without a ChatGPT login and should support low-bandwidth use and local-language instructions.

Store role, chosen discipline, attempts, consent status, task version, score components, human-review decision, certificate status and any accepted knowledge object ID. Keep public demonstration analytics separate from the candidate’s identifiable record.

7 Pilot acceptance and metrics

Acceptance: four working synthetic demonstrations; diagnostic and final task in at least two tracks; all four qualification classes represented in the rule model; one human appeal; one consent withdrawal; one candidate result routed to a scholarship or work review without automatic award; one reviewed student contribution that enters the knowledge queue; and one company demo that creates a scoped lead.

Measure completion, qualified enrolment, accepted tasks, correction rate, time to competence, consented candidate-to-paid-work conversion, demo-to-pilot conversion and repeat service. Compare cohorts by language and site. The go/no-go decision for expansion rests on accepted quality and cost, not registrations alone.

8 Ownership and staged release

The central education team owns curricula, assessments and credential rules. Local experts validate regional facts and supervise practice. Product engineering owns the game, identity, analytics and interfaces. A rights officer approves reuse. The Guinea experimental centre is the first integration and field-test site; the UAE head centre controls released packages. Stage gates are content and rights inventory, playable prototype, supervised pilot, independent quality review and regional deployment.

9 Cross-system training and capability map

The development matrix shall carry two cross-cutting assignments against each applicable chapter: knowledge and skill registration, and education and candidate selection. These assignments indicate where a chapter contributes evidence, a reusable module, a practical case or an assessed competence. They do not change the chapter's existing development status or imply that the proposed service is deployed.

Each course task shall identify the system function it teaches, the source data and permissions it uses, the human qualification it tests and any separately tested agent capability. Students move through Class III, Class II, Class I and Expert within a discipline; agents use capability passports rather than human certificates. Supervised results may become candidate knowledge objects only after rights clearance, source verification and specialist acceptance.

The same funnel shall record client demonstrations, scoped enquiries, paid pilots, repeat assignments and qualified local implementers as separate conversion stages. Candidate selection, scholarship review and contributor payment require published criteria and consent. A verified contribution can be linked to an information or skill module for later use and remuneration; training participation alone grants neither a licence to the work nor an automatic financial valuation.

References

Source basis: Oristella learning-system dialogue of 21 October 2026; Oristella Information Tokenisation Proposal EN of 20 October 2026; current six-appendix investment proposal. Requirements below are proposed design decisions, not claims of deployed functionality.

Credential format: W3C Verifiable Credentials Data Model 2.0 https://www.w3.org/TR/vc-data-model-2.0/ . The certificate is an evidence record; employment and course accreditation require separate agreements.

Related technical assignment: Oristella Regional Knowledge and Skill Registry, section 8. The chapter-level cross-cutting map is maintained in the development matrix.

← Return to main appendix