21 October 2026. Six sections and six appendices; the financial model is attached as two Excel files (Appendix 6.1).
Oristella began as the AI that runs a gold mining company. It has become the AI that builds gold mining companies.
1 Strategy
Expertise within reach, at a substantially lower entry cost. A junior gold company in Guinea holds a licence, a deadline and no geologist on staff. Consultants quote six figures and four months. Oristella delivers the package at a fraction of the cost and time, with experience gained at Farafina Gold Group.
Industry structure. Small operators have limited bargaining power against consultants and depend on specialist software and scarce expertise. Shared infrastructure and reusable workflows change that cost structure.
Strategic position. A niche player with absolute focus on gold mining, combining lower delivery costs with specialist quality. Oristella supplies documentation and management capabilities: its agents act within defined decision rights, with named responsibility for accepted results.
Value proposition. A defined job at a fixed price. The client receives the output of a full expert team without building one; later updates are expected to be far faster and cheaper than the first delivery.
Barriers to entry. Oristella lowers the two barriers that keep clients out of mining - cost and time - and raises the barriers to entry into its own market for competitors: accumulated knowledge, the experience curve and a regional physical base.
Sustainable competitive advantage. The lasting asset is accumulated knowledge: enterprise models, decision history, agent skills. Tools can be swapped; the knowledge stays. Documented regional requirements add context that general models do not receive by default. Retention rests on accepted results, accumulated context and integration, while clients retain access to their own data.
Demonstrate, then contract. The starting market is all of Guinea; Sierra Leone and the neighbouring countries follow. Initial requests are already in hand: licence-extension documentation, geological modelling and mine planning.
Route to the client. A trusted introduction, a live demonstration at the Guinea data centre, a fixed-scope paid assignment. Farafina Gold Group is the working showroom for nearby companies.
Delivery experience. Three staff prepared the Lions Head Resources licence-extension package in three weeks, against an external quote of about $300,000 and four months; the ministry approved it. Two related-company assignments, for Farafina Resources and Lions Head Resources, provide the initial workload on the same terms as for all clients; two external orders are in preparation, with contracting planned after on-site equipment testing.
Regional cluster. Regional centres operate under an Oristella franchise: the partner funds premises, equipment and deployment; Oristella supplies software, knowledge, training and support. A second centre, in Sierra Leone, is in preparation for 2027.
Education as a market instrument. Product-linked programmes with local universities build a talent pool, customer loyalty and a route to orders. The proposed joint diploma "Certified Oristella Operator" is established through partner agreements; trained operators reduce the staffing needed for new clients.
Four months to a testable first release. Ori 1.0 - documentation, mapping, geological modelling, drilling and mine planning, plus the first data centre - targets a testable release within four months of funded deployment. Ori 2.0 - production, finance and transport management - follows within the year. Formation of Oristella Ltd in a Dubai free zone is paid for and under way.
Work distributed by reaction time. The central complex handles tasks measured in days; the local data centre handles tasks measured in hours; computers at extraction points work online.
Quality management and automation. Every result and the work it saved are measured. Routine work, from document preparation to checking compliance with local mining law, is expected to be largely automated: GPT prepares, Claude checks, specialists approve.
One information loop from fact to action. Observation, validation, model, decision, execution, feedback: one data model carries every step. Central models, local AI on site and specialist agents form three levels of one network. A matrix of 27 sections and 202 chapters tracks readiness.
Data and continuity. Data reside locally, on Oristella servers, or both; client records are kept separate. Knowledge survives changes of people, software and AI models.
Security. Every consequential action is traceable: client data separated, approvals recorded and safety-critical equipment kept within approved local control systems. Continuous checks by specialised AI agents are part of the system design.
Development direction. Ori 2.0 and Ori 3.0 are designed to register information blocks and agent skills with their source, rights, use and contribution to accepted work; a controlled token register will support exchange and contributor remuneration between centres.
People and agents with defined authority. Management is one system of experts and AI agents, each with a job description, rights and a record of actions. Routine decisions follow delegated authority; exceptions and conflicts escalate to the accountable manager.
Labour substitution and expert leverage. Automation removes routine preparation, searching, formatting and data handling, so scarce experts spend their time on judgement. Planning assumes 50-60% automation of defined information tasks; higher rates are confirmed by measurement.
Team. System design and local delivery are led by the people who built the management systems at Farafina Gold Group.
Design: Vasiliy Maltsev, Chief Designer, head of development
Natalya Barytskaya, Doctor of Sciences (Geology)
Viktor Bagenov, mining engineer
Evgeny Gurevich, AI and business administration consultant
Alexey Kabernyk, metallurgical engineer
Ilya Karas, Chief Scientific Adviser (systems analyst by training)
Data Centre: Ousmane Tunkara, Director, Data Centre
Funding the commercial launch. The founders make an operating base available, valued by management at about $460,000 (land, buildings, equipment, software and prior development). Oristella seeks $460,000 in cash to complete deployment, commissioning, product development and working capital. The planned tranches are $300,000 in November 2026, $110,000 in January 2027 and $50,000 in April 2027. Oristella Ltd is being formed in Dubai; until the fund approves the budget, the founders perform its role, and in Guinea the work is carried out by a founders' group company directed by Ousmane Tunkara.
Forecast. The base case projects revenue of $997,200 and profit after tax of $398,078 for FY1 (November 2026-December 2027, 14 months), followed by revenue of $2,043,061 and profit of $1,327,343 in FY2 (2028). The forecast assumes 10% quarterly revenue growth in 2028 and flat monthly cash operating costs. Separately contracted expert fees and physical fieldwork are outside Oristella revenue. Flat operating costs and one expert supporting three to five companies are planning assumptions, validated against workload and review hours.
Cash and resilience. Base-case closing cash is $514,233 at the end of FY1 and $1,792,023 at the end of FY2. With customer receipts 20% lower and cash operating costs 10% higher, it is $236,397 and $1,056,979. The lowest stress-case month closes with $42,330 in December 2026. Under the stated receipts, funding and cost assumptions, neither case requires additional financing; no distributions are assumed. The fund will propose the investment instrument and terms.
Appendix 6.1 Financial Model: Base Case and Stress Case Balance Sheets and Cash Flows (two Excel attachments: Oristella_Base_Balance.xlsx, Oristella_Stress_Balance.xlsx)
Oristella occupies a defined niche in services for junior gold mining companies, combining lower delivery costs with specialist service quality. It expands its market by lowering the cost and time required to prepare projects. Accumulated knowledge, integrated workflows and local delivery strengthen retention and raise the effort required for competitors to reproduce the service. For investors, the objective is a scalable business with recurring revenue and growing cash generation.
Oristella supplies cognitive capabilities: specialised information, the skill to process it and completed production tasks with an accepted result. The first documentation service proves the method; the system’s value grows as these capabilities are applied across companies and regions.
Strategic principles
1. Competitive Position. Focused differentiation supported by a lower-cost delivery system. The initial market is junior gold mining companies in Guinea, followed by neighbouring countries. This narrow focus develops expertise beyond the reach of generalist providers. The combination of differentiation and low cost holds because AI agents shift the productivity frontier: the same accepted result costs less and takes less time. The frontier shift alone can be imitated; the position is held by the activity system and accumulated regional knowledge.
2. Strategic Objective. Build the knowledge infrastructure for junior gold mining. The scope grows from document preparation to enterprise management, while deployment extends from one site to a regional network. At each stage, the system takes on a greater share of the client's operations.
3. Experience Curve. Planning starts with 50-60% automation of defined information tasks. Higher substitution rates and one expert overseeing four or five systems are task-specific targets, measured against accepted work, review hours and correction rates. Validated lessons from completed assignments improve the next delivery.
4. Proprietary Capability. The system's cognitive capabilities are developed within Oristella as an integrated specialist capability. Labour substitution is measured separately for each profession. The level of automation is set to maximise economic returns, taking account of the point at which marginal benefits begin to decline.
5. Revenue Model. Clients pay a fixed fee for each assignment and recurring licence fees thereafter. Licence fees increase with the number of modules deployed and the volume of work performed for each client.
6. Learning Effects Across Sites. Each operating site can contribute authorised, validated knowledge through the central system. Reusable methods and regional findings increase the value of the service across clients. The commercial effect is measured through repeat use, delivery costs and accepted results.
7. Market Expansion. Lower entry barriers make smaller deposits commercially viable and create demand from new clients. Multilingual access opens further markets, while franchised centres extend the business into new regions.
8. Customer Retention. Accumulated context, service quality and integrated workflows make continued use valuable. Client data remain accessible to their owners.
9. Barriers to Entry. Accumulated experience, learning effects across sites, access to universities and government support make entry harder for competitors. Each additional client strengthens these barriers.
10. Operating Leverage and Cash Generation. Reusable software and knowledge reduce the incremental effort per assignment. The base forecast retains cash for development and selective expansion. Further automation of expert assessment is a priority; the forecast does not assume additional savings before they are demonstrated.
11. Client Pipeline. Each module is accompanied by courses at universities specialising in the relevant disciplines. These courses train the workforce and generate client enquiries. Graduates operate Oristella for clients and at regional centres.
12. Risk Management. Each module is tested and refined under actual operating conditions at Farafina before release to external clients.
13. Core Competence. Experience in managing gold mining operations in Africa. This expertise is embedded in the system's algorithms and standards.
14. Intangible Asset. Knowledge is Oristella's principal strategic asset. Ori 2.0 and Ori 3.0 develop a register of information, specialist capabilities, rights, use and remuneration. Internal economic value guides development priorities; financial asset recognition follows the applicable accounting criteria. Tokenisation alone does not create a balance-sheet value.
All other sections put this strategic position into practice.
Strategic discipline
The position rests on a narrow gold-sector focus, a distinct delivery system and mutually reinforcing activities: reusable information contracts, specialist review, local support and product-specific training. Together they produce lower cost per accepted assignment and sustained client value - Porter's activity-system logic applied to a concrete business. [P1, P2]
The first market offer is documentation, modelling and recurring updates. Expansion follows evidence of repeat orders and delivery margins. Financial recognition of knowledge follows the policy in Appendix 6, while operational tokenisation is developed in Appendix 4.
The entry market is small gold mining and exploration companies in Guinea with an active project, usable source data, a near-term obligation and the capacity to pay. Licence counts define the prospecting pool; paying demand is confirmed client by client. Documentation, geological models, exploration plans and project reporting are the first offer because they address defined tasks already represented in the project materials.
Evidence of the addressable groups
Country and date
Gold licence data (published sources)
Guinea14 May 2025 (revocation decree)
26 grouped holder names and 31 distinct published revocation records from one event; a reference for the prospecting pool, not an annual total. [M1]
Sierra Leone31 December 2023
17 holders of 17 small-scale gold licences; 2 holders of 2 exploration licences. One holder overlaps. [M2]
SenegalFirst half 2025
75 normalised holder names across 107 small and semi-industrial Au authorisation records. [M3]
GhanaDecember 2024
82 small-scale gold licence holders in the Eastern Region, reported from the registry by Hilson and colleagues. [M4]
LiberiaDecember 2024
84 gold Class B licence holders, reported from the registry by the same researchers. [M4]
These groups differ in geography, licence class and date and are shown separately, without a combined regional total. Documentation quality is diagnosed client by client. [M1, M5]
Entry and recurring service
The sales sequence is a trusted introduction, an authorised demonstration, a defined paid assignment and acceptance. Oristella module tariffs in Appendix 6.1 are $1,000 for a trial geological model, $5,000 for a full model and $300 for a standard update. These are Oristella revenues, excluding separately contracted experts and physical fieldwork. Each contract defines scope and module composition. The commercial target is two related-company clients from January 2027, two external clients from April, and one new client a month from May to December, reaching twelve by December 2027. Revenue is forecast by service stream; the client target does not independently determine receipts.
For comparable work, the commercial target is a total client price 1.5-2 times lower than traditional delivery and a delivery period two to three times shorter, subject to scope and source-data quality. Initial pricing allows 50% of the combined professional-service fee for Oristella and 50% for experts. Thus an Oristella fee of $5,000 for a full model implies a combined fee of $10,000 at that split. On highly automated tasks, the expert share may fall to 5-10%; the total fee is then the Oristella fee divided by 0.95-0.90. These are pricing assumptions, not percentages of expert time. Physical field execution is provided and priced by partners. Standard updates reuse accepted inputs; materially new work is quoted separately.
Initial assignments. Expert assessment is performed within the system. For the first two external orders, it is included in initial delivery costs with no separate assessment fee. The assessment workflow is a priority for automation; subsequent pricing reflects the specialist work actually required.
Three staff prepared the Lions Head Resources licence-extension package in three weeks, against an external quote of about $300,000 and four months; the ministry approved it. Two related-company assignments provide the initial workload on the same terms as for all clients; two external orders are in preparation, with contracting planned after on-site equipment testing.
After acceptance, Oristella offers model updates, recurring reporting and further modules. A $300 standard update is 6% of the $5,000 model tariff; materially new work is agreed separately. Sierra Leone follows through a proposed second regional centre, with discussions and agreement preparation under way for 2027. Further countries follow qualified demand. Franchise partners finance local infrastructure; Oristella prices software, knowledge, training and support separately. The current model includes no incremental franchise income.
Two connected entry funnels
For clients, a free interactive demonstration on synthetic data shows a draft geological, financial or documentation result and leads to a fixed-scope paid assignment. For students and specialists, a short gold-company management game leads to a diagnostic task, a discipline course, a supervised examination and, for selected candidates, paid work. Approved regional assignments become course cases; verified results enter the knowledge base after expert validation. The Guinea pilot measures conversion from demonstration to paid assignment and from first delivery to repeat service, and the number of assessed learners ready for supervised work.
The first installation is Oristella's permanent test, demonstration and commissioning data centre in Guinea. It tests the system under real operating conditions with physical equipment connected, demonstrates the service to prospective clients and trains operators. Oristella-owned additions and Farafina-provided assets or access rights are recorded separately.
Farafina Gold Group makes the existing operating base available to the founders for the first centre. Premises and land are allocated, buildings fitted out, equipment delivered, and electricity and air conditioning provided; the site is ready for installation and staff accommodation. Management estimates the combined founder contribution at approximately $460,000, including developed software, models and prior work. The cash funding completes deployment. An inventory will distinguish the provided resources from additional purchases.
Capacity as a service. The centre is designed for testing, demonstration and difficult integrations beyond the company's immediate production requirements. Oristella can use that capacity to deliver outsourced processing and support to Farafina and external clients. Each service requires an agreed scope, price, resource allocation and responsibility for the result; testing and existing commitments take priority.
Installation and acceptance
The installation models two-level interaction between a local centre and the central server: local knowledge bases, information exchange, specialist applications, physical interfaces and workflows with large AI models. Tests measure output quality, processing time, resource use, operation during connectivity outages, synchronisation after reconnection and recovery from faults. Each test records the software build, equipment configuration, input data, responsible specialist, acceptance criteria and result; replication follows recorded acceptance.
Release sequence and professional control
Ori 1.0 - documentation, mapping, geological modelling, drilling and mine planning - is released within four months, with further functions within the year; each release follows installation, integration and acceptance. GPT-5, GPT-6 and Google Gemini support design; Claude cross-checks and audits; DeepSeek is used selectively. Responsible specialists approve consequential outputs and provide required professional signatures.
Planning assumes conservatively that 50-60% of defined information tasks are automated, measured on accepted assignments including specialist review, corrections and staff time. Higher task-level rates, up to 95%, are confirmed against their own measured baselines.
Central complex and franchise deployment
A separate central complex in the United Arab Emirates coordinates software releases, common knowledge resources and specialist processing for the regional network. The Guinea centre continues to test updates, demonstrate applications and develop non-standard hardware, software and integration configurations.
Franchise partners provide and fund subsequent local premises, equipment and physical installation. Oristella supplies software complexes, structured knowledge bases, training, updates and support. Each centre follows an approved configuration with responsibilities for licences, maintenance and acceptance. This avoids Oristella capital investment in partner infrastructure; software delivery, specialist processing and support still have costs. The financial model funds the first Guinea centre. No separate expansion budget or franchise receipts are added to this base case.
Oristella integrates AI, specialist applications and information management for small gold mining and exploration companies. Its initial scope covers documentation, mapping, geological modelling, drilling, samples and laboratory data, with links to planning and management.
Technical requirements and outline design
Each object has an identifier, passport, source, owner, version, quality state and change history. Processing preserves raw inputs and source relationships. Adapters connect specialist applications to the common model. Exchange contracts govern receipt, validation, transformation, errors and quarantine. Significant actions require authorisation and leave an audit trail.
The target architecture connects client equipment, regional data centres and the separate Oristella central complex in the United Arab Emirates. Local centres retain knowledge bases and essential processing capabilities; the permanent Guinea centre validates the interaction between a local centre and the central server before configurations are released for wider deployment.
Operational use and standardisation
Google Gemini, automatic mapping, automated sample-preparation accounting, geological modelling at a level comparable to Micromine and the existing financial system are in active use; Odoo is being prepared for implementation. Standardised documentation, mapping and modelling have shown high practical effectiveness and underpin the first deployment stage. These solutions are now being standardised for replication and integration.
Security and intellectual property
Client data remain separated; access follows role and task. Shared libraries contain only material Oristella may reuse. Safety-critical equipment remains within approved local control systems. Persistent module identity, provenance and licence history support reuse and payment accounting.
Information register and contributor remuneration
The register distinguishes raw information, validated datasets, work in progress, software, AI modules and released deliverables. Each object records its version, provenance, rights holder and permitted use. A non-exclusive licence gives the customer a usable copy while Oristella retains rights to its underlying module; records identify the recipient, terms, payment and contributor share. The proposed internal token system records authorised use and settlement between centres; clients pay for contracted results. Each new assignment can add validated regional knowledge. Economic value is assessed through accepted use, avoided work and the contribution to delivery quality.
A permissioned blockchain register tracks module identities, licences, authorised use and contributor payments; confidential content stays in controlled storage. Non-fungible tokens can identify modules or licences, and an internal Oristella token serves as a tracking unit without an initial market price. Each licensed release carries a signed hash of its canonical version, a per-recipient issue record and a copy-specific fingerprint, which support investigation of unauthorised redistribution. The register complements financial accounting; tokenisation by itself neither establishes asset value nor transfers copyright, and the forecast includes no token revenue. [A1, T1, T2]
Above general AI models, Oristella builds a controlled regional evidence layer. Local centres collect authoritative rules, accepted document formats, project observations and community requirements. Each record identifies its source, jurisdiction, date, status and reviewer. Ministry comments and regional preferences are retained as dated project evidence, separately from binding legal requirements. For example, a processing-capacity preference raised during a review is not treated as a universal licence threshold. Lessons, objections and corrections enter a restricted internal corpus, then validated functions and agent skills. Only approved, rights-cleared releases become licensed knowledge packages; student findings require permission and expert validation.
Each regional data centre maintains local knowledge bases and the software for its authorised work; the central complex coordinates common software, model and knowledge releases, shared methods and specialist processing. Client and site records keep their own ownership, permissions and change histories, and only information authorised for transfer or reuse enters shared resources.
Processing and exchange under limited connectivity
The engineering reference assumes a symmetric 10 Mbit/s site connection and a remote connection of up to 100 Mbit/s; the site link remains the transfer bottleneck. Raw video, essential local services and immediate equipment interfaces remain on site. Selected records, events, asynchronous jobs and approved updates pass between centres. Transfer priorities and daily data volumes must be tested together.
Variant A places Micromine and Metashape locally and shares the graphics processor between engineering and AI workloads. Variant B runs the engineering applications remotely and keeps local graphics capacity for AI, subject to licence rights and measured connectivity. Commissioning selects the variant and defines the minimum uninterrupted local service set, resource limits and recovery procedures.
Information passports, exchange forms and protocols govern object identity, ownership, versions, validation and state transitions. Working design names the authoritative writer for each object and specifies delivery acknowledgement, duplicate handling and reconciliation after an outage.
Current work and development status
Oristella is standardising and integrating operational applications for wider deployment: information exchange, software configuration, trials and operator training. The Guinea site is ready for installation and personnel.
Position at 21 October 2026.
Current work programme
Workstream
Current position
Next work and acceptance milestone
Documentation and modelling
Documentation, mapping, sample-preparation records, Google Gemini and geological modelling comparable to Micromine are in practical use.
Standardise packages and demonstrations; validate representative assignments for wider deployment.
Information flows
Passports, processing states, forms and protocols documented: 40 interfaces and a 53-sheet geological interface album.
Close design comments; complete schemas and adapters; test exchange, validation, traceability and recovery.
AI and applications
GPT-5, GPT-6 and Google Gemini support design; Claude reviews and audits; DeepSeek is used selectively. Finance software is in use; Odoo deployment is being prepared.
Confirm versions, complete integration and test end-to-end workflows with specialist approval.
Guinea data centre
Premises and land allocated; buildings fitted out; equipment delivered; electricity and air conditioning provided. Ready for installation and personnel.
Reconcile the equipment and licence inventory; install and configure the system; test connectivity and workflows.
People and market preparation
Product-linked training and delivery roles defined. Initial services use established documentation and modelling methods.
Complete demonstrations and courses; assess operators and deliver a supervised client pilot.
Information and cognitive capability registry
Concept and technical assignment prepared. The design covers information identity, provenance, rights, licences, use, remuneration and versioned agent skills. Operational blockchain and asset recognition follow the pilot.
Inventory rights and sources; pilot a permissioned register with protected off-chain content; test offline sync, licence accounting, negative knowledge and accounting reconciliation.
Learning programmes, talent reserve and agent capability growth
Four-class qualification ladder, student and client entry journeys, practical regional cases and assessment rules specified. Demonstration model drafted; production courses, credentials and recruitment follow the pilot.
Test a synthetic-data game and company demonstration; pilot two disciplines in Guinea; assess humans and agents separately; review thesis awards, talent-pool consent and controlled agent releases.
Design maturity and trials
The matrix covers 27 sections and 202 chapters. Information flows include detailed working-design material. Integration is in technical design and programming; operational applications are being prepared for replication and trials. The sequence is configuration, installation, interface and workflow tests, then a supervised client pilot. Debugging and acceptance will be recorded against identified software builds.
Two further workstreams have their own technical assignments: the controlled information and cognitive capability registry, and the learning system for local recruitment, qualifications and agent improvement. Each is approved separately with its own timetable, budget and commercial effect.
Distributed Data Centre Architecture and Deployment
Engineering basis
The engineering basis is Data Centre Structure v1.0 (October 2026, Russian-language reference available on request): an indicative workstation with a 96 GB graphics processor, network storage, networking, backup power, selected cameras, trackers and an equipment test rig. Configuration and prices are preliminary engineering assumptions; latency, availability, concurrent workloads, licence rights and installed equipment are confirmed in commissioning.
Working design and test records
Detailed releases identify the equipment and licence inventory, deployment diagrams, software versions, local knowledge-base structures, information passports, exchange forms and protocols, each with its revision and responsible reviewer.
Commissioning extends the tests in Appendix 3 to resource contention, queued transfers, data reconciliation, access separation, backup and restoration. AI models are assessed on representative tasks rather than nominal size.
Engineering decisions before replication
The minimum uninterrupted local service set is reconciled with graphics-processor sharing in Variant A. Data-transfer schedules cover simultaneous photography, backups, telemetry and interactive sessions. Variant B is validated for licences and remote sessions. Recovery design specifies acceptable data loss, restoration time and return to normal operation, and the capacity of the central complex covers concurrent regional demand.
The Excel model is the investment baseline for the first Guinea centre. The engineering estimate is an alternative planning scope, not an addition to the budget, and is reconciled line by line with the model before procurement.
Appendix 5 Human and AI Resources Policy
Oristella combines specialists, local operators and AI agents in one system of responsibility. Each agent receives a defined task, tools and permissions and preserves the provenance of its work. The responsible specialist checks quality and makes the professional decision. Client specialists retain their professional responsibilities as routine work is automated.
Role descriptions specify purpose, inputs, expected output, read and write rights, prohibitions, escalation and accountability. Agent passports and human responsibilities must agree. Changes to another functional domain’s accepted state follow its authorised procedure. Significant actions and approvals are recorded; conflicts escalate to the accountable manager.
Training and qualification
Each delivered software package is accompanied by a centrally maintained online course tied to its version. The Oristella AI tutor checks exercises and assessments; individual human curation is offered to the most capable learners or for a fee, and in-person sessions to the most gifted. The central system maintains assessment rules, versions and results; approved local tools support practical work during interruptions in connectivity, with subsequent synchronisation.
Initial learning covers system operation and maintenance, input preparation, geological and mapping tasks, checking and issuing documents. Later streams cover marketing, financial analysis and operational management. Learners may qualify in several fields. Assessment requires a reproducible data package, explanation of sources, correction of errors and respect for access rights. Regulated technical conclusions remain signed by licensed professionals.
University geology faculties, professional associations and consultants are potential delivery partners; the joint diploma "Certified Oristella Operator" and other certification are established by formal agreements with them. Qualified learners may be offered suitable assignments, including paid information processing as orders arrive. The commercial purpose is a local support base that lowers implementation effort and improves sustained use.
Team and capacity
Design: Vasiliy Maltsev, Chief Designer, head of development
Natalya Barytskaya, Doctor of Sciences (Geology)
Viktor Bagenov, mining engineer
Evgeny Gurevich, AI and business administration consultant
Alexey Kabernyk, metallurgical engineer
Ilya Karas, Chief Scientific Adviser (systems analyst by training)
Data Centre: Ousmane Tunkara, Director, Data Centre
Maltsev previously headed Guinea Geological Research, which developed management systems for Farafina Gold Group.
Tunkara previously headed Gelecs SARL (Guinea), working on technical systems, engineering applications and aerial surveying, and leads practical implementation in Guinea. This continuity links system design with local delivery.
Capacity is planned through named roles, availability and compensation consistent with the financial model. One expert supervising four or five systems is validated against review hours and service commitments; automation and workload-reduction rates are measured task by task.
Scaling follows signed role commitments, responsibility for each accepted output, workload limits, substitute reviewers and costed training delivery. Tracked indicators are accepted assignments, correction rates, time to competence, expert utilisation and subsequent paid use. New centres and training programmes carry their own staffing and cost calculation.
Initial training programme
The proposed first programme follows the first deployed service package. Each stream ends with a supervised practical assignment and a record of competence. Detailed teaching hours and examination materials follow the selected software configuration.
Stream
Practical outcome
System operation
Configure an authorised workplace; manage access, local work, synchronisation and escalation.
Information preparation
Create a source-linked dataset; complete object passports and resolve validation errors.
Geology and mapping
Prepare and review a reproducible model or map using an approved workflow.
Documentation and service
Produce a checked client package, document changes and process a subsequent update.
Financial and operational analysis
Use the relevant modules, reconcile outputs and recognise decisions requiring specialist approval.
Client development
Qualify a request, demonstrate a defined service and hand over an agreed assignment.
The Oristella AI tutor assesses practical work; local centres host supervised in-person sessions where needed. The central team maintains versions, course content and certification rules. Qualified participants may contribute to paid information work when assignments exist; contributor rights and payments link to the information register in Appendix 4. University partnerships and public communication support local adoption, subject to agreed responsibilities and recognition.
Four qualification classes and a live learning funnel
A free university game on a synthetic gold-company case introduces students to Oristella through geology, finance, market analysis and documentation; a parallel client demonstration shows a draft result on training data before a scoped paid pilot. Both routes reach real tasks only under rights, confidentiality and specialist-review controls.
Each discipline has four qualification classes - Class III, Class II, Class I and Expert. Entry diagnostics, guided cases and independently assessed final work determine progression and delegated authority. Strong candidates may receive scholarships, thesis awards, paid information work and employment under published criteria, with their consent and as positions open. Accepted regional findings and documented failures enter a reviewed knowledge queue; human experts approve every production release, and AI agents have their own task passports and test records.
Appendix 6 Financial Strategy and Ask
Oristella requests $460,000 to complete the first permanent Guinea testing and demonstration centre and launch the service business. The founders make an existing operating base available, valued by management at about $460,000. This includes land, buildings, utilities, accommodation, equipment, software, models and prior development; Dubai formation costs have already been paid. Oristella Ltd is being formed in Dubai; until the fund approves the budget, the founders perform its role, and in Guinea the work is carried out by a founders' group company directed by Ousmane Tunkara. Organisational use of the funds is settled after the fund's approval. Internal orders support the starting workload and are included in the operating forecast rather than added again as an asset or financing receipt.
The cash request is staged: $300,000 in November 2026, $110,000 in January 2027 and $50,000 in April 2027. Existing purchase budgets provide $90,000 for additional equipment and $69,000 for software rights; the remaining $301,000 supports commissioning, development and working capital. The forecast retains the funds and contains no scheduled repayment or distributions.
Two financial cases cover FY1 (November 2026-December 2027, 14 months) and FY2 (2028). The base case assumes the source tariffs, delivery schedule, 10% revenue growth at the start of each quarter in 2028, and flat monthly cash operating costs. Revenue is $997,200 and $2,043,061; profit after tax is $398,078 and $1,327,343. The figures cover Oristella services and do not add uncontracted franchise, outsourcing or token revenue. Separately contracted expert fees are outside Oristella revenue. Flat cash operating costs and one expert supporting three to five companies are planning assumptions subject to workload and review hours.
The stress case reduces customer receipts and matching sales by 20% in every forecast month and increases cash operating costs by 10%. This tests reduced incoming operating cash, rather than imposing a fixed percentage reduction on accounting profit. The financing, deployment purchases and existing development-capitalisation schedule are the same in both cases. No additional financing is requested.
Why the funding is needed. In the stress case, cash reaches its lowest point of $42,330 in December 2026. The company uses financing during deployment and the first sales cycle: free cash flow is negative $223,603 in the first reporting period. Closing cash recovers to $236,397 at the end of FY1 and $1,056,979 at the end of FY2, assuming the scheduled receipts and funding. Delayed receipts or higher deployment costs would reduce this small early cash cushion.
USD
Base FY1
Stress FY1
Base FY2
Stress FY2
Revenue
997,200
797,760
2,043,061
1,634,449
Profit after tax
398,078
145,246
1,327,343
888,530
Net operating cash
464,738
186,902
1,501,350
1,044,143
Free cash flow
54,233
(223,603)
1,277,790
820,583
Closing cash
514,233
236,397
1,792,023
1,056,979
Minimum monthly cash
51,300
42,330
599,640
286,322
Additional financing
0
0
0
0
The base case generates free cash flow of $54,233 in FY1 and $1,277,790 in FY2; the stress case generates negative $223,603 and positive $820,583. Cash retained at year end includes unused financing and remains available for operations and development. These balances are not an automatic investor distribution.
All amounts are in US dollars. Appendix 6.1 contains two Excel files. Each opens with its balance and cash flow, supported by editable assumptions and the monthly calculation. Both balances reconcile throughout the forecast.
The financial model provides for the first Guinea centre as the permanent experimental and demonstration base, including physical equipment connections, commissioning, software testing and work with large AI models. Its planned capacity exceeds Oristella internal needs and may support paid outsourced processing for Farafina and other clients under separate contracts. Such receipts are a potential additional revenue stream; they are not added to the present forecast. Subsequent centres are planned on partner-funded franchise terms. The separate UAE central complex does not replace or relocate the Guinea demonstration centre.
Budget reconciliation. The $159,000 of additional equipment and software purchases remains in the forecast and must be reconciled with the provided inventory before procurement. Existing site and electricity costs are retained. Founder-provided resources do not create a cash receipt; no founder cash injection is assumed.
Basis of the balances. These are management forecasts for the Oristella development project. Founder resources have a management memo value of about $460,000. The forecast assumes no cash purchase or transfer of title and excludes that memo value from assets and equity. Statutory recognition follows the executed agreements and applicable standards. Development capitalisation is a planning assumption of 45% of development-team costs; qualifying expenditure and the date software is available for use require project records. Corporate tax is modelled at 9% above AED 375,000; no 0% free-zone rate is assumed.
Expert assessment. Assessment is performed within the system and is a priority for further automation. For the first two external assignments, no separate assessment fee is charged; the work is included in initial delivery and development costs. Subsequent costs fall as repeatable tasks are automated. The forecast includes no additional cost saving from that future automation.
Financing terms. The fund will propose the investment instrument and terms. This package presents the investment proposal, the use of funds and two financial cases.
Appendix 6.1 Financial Models
The editable Excel files contain the complete base and stress scenarios. FY1 covers November 2026-December 2027; FY2 covers 2028. The views below reproduce the saved workbook results.
