Methodology, published.
Score, reconciled.

Afrimintel was developed through April 2026 under the earlier working name STRATA. The platform is now launching publicly as Afrimintel. Audit log entries and technical reports dated before this release retain the STRATA reference as historical record.

The problem Afrimintel is built to solve

African mineral assets carry a verification discount when they reach global capital. A DFI credit committee, a mining-finance desk, or a sovereign-wealth investment team applies a structural haircut to African geological data that does not exist for comparable data from Canada, Australia, or Chile. The discount is not a judgement on the geology. It reflects thinner verification chains — fewer historical NI 43-101 / JORC sign-offs, harder-to-reach primary sources, higher opacity on methodology.

Afrimintel is built to close that gap. Not cryptographically — no blockchain, no verifiable credentials, no distributed ledger. Those solve a different trust problem. Afrimintel closes the gap methodologically:

• Source hierarchy. Every data point is tagged to its source tier: primary regulatory filing, primary academic source, secondary institutional, or tertiary commentary. Lower tiers are visible but marked.
• Epistemic tagging. Every factual claim carries a Sourced, Derived, or Absent tag. If a number is an inference, the reader sees it.
• Public audit log. Every correction since April 2026 is published at /audit-log/, filterable by type and scope. No correction is buried. Most commercial mineral intelligence platforms do not publish their errors; Afrimintel does.
• Three-tier deposit classification. Decision-grade, scored-reference, spatial-reference only. The distinction is disclosed per record.
• Known Gaps and Verdict Reversal. For each of 13 geological provinces, Afrimintel publishes what is missing from the current intelligence base and the conditions under which the current verdict would reverse.

This is the methodological claim. The test is whether it holds under challenge. The Corrections and Feedback section below carries a direct email for errors found by readers. If you find one, use it.

What Afrimintel is

Afrimintel is an Africa-focused mineral intelligence platform covering 13 geological provinces and 40 countries. Current data inventory:

• A deliberately curated deposit corpus across the continent (distinct assets; near-duplicate entries retired June 2026)
• 18 intelligence-grade records — carry resource tonnage, grade, operator, and primary-source citation (NI 43-101, JORC, or audited annual report)
• Research-grade records — sourced from published company or technical disclosures, carrying published figures — resource grades and tonnages, or, for discovery-stage assets, drill results — in narrative form, but not yet migrated to the structured parameter tables (and, in some cases, the full primary-source citation chain) that define the intelligence-grade tier
• Operator-attestation records — sourced from named-operator disclosure but without a QP-signed NI 43-101 / JORC framework
• Spatial-reference records — carry name, coordinates, commodity, status, and province only (no resource dossier)
• 19 AI tools across 5 professional roles
• 19 free data source references

Afrimintel is currently in closed beta. No paid tier is available for purchase.

Continental policy framework context

Afrimintel operates within — but does not claim formal alignment with or endorsement by — the recognised continental policy architecture for African mineral resources. Where work by Afrimintel touches questions these frameworks address, the framework is acknowledged.

• Africa Mining Vision (AMV) — adopted by African Heads of State at the February 2009 African Union Summit; the foundational continental framework for mineral-sector governance and value addition. Implementation has been mixed across African Union member states; multiple academic and civil-society assessments since 2015 document significant implementation gaps relative to the framework's ambitions. Afrimintel's methodology touches AMV principles through the Country Risk Composite's resource-governance dimension — the NRGI Resource Governance Index component, weighted at 25% within the composite — and through the disclosed scoring of provincial governance dimensions.
• African Minerals Governance Framework (AMGF) — adopted by the African Union Ministers of Mineral Resources Development in 2016 as the policy instrument for monitoring AMV implementation across African Union member states. Developed through the African Minerals Development Centre (AMDC), originally hosted at UNECA; AMDC was officially handed over to the AU in February 2019 and is now a specialized AU agency. Afrimintel's methodology adopts AMGF-style monitoring discipline through its quarterly deep audit cycle on rotating provinces and its public audit log of corrections.
• African Green Minerals Strategy (AGMS) — crafted by the African Development Bank in 2023, adopted by the African Union at the 38th Ordinary Session of the AU Assembly in February 2025. Articulates Africa's positioning in green mineral (cobalt, lithium, copper, manganese, nickel, graphite, rare earth elements) value chains amid the global energy transition; argues for downstream value retention as exit from the primary commodity trap. Afrimintel's Critical & Battery Minerals sub-vertical references AGMS as the continental policy framework for African green mineral positioning; the sub-vertical's province profiles surface the green mineral commodities AGMS prioritises.

Acknowledgement is not endorsement and is not a claim of partnership. These frameworks are part of the institutional landscape Afrimintel operates within. Implementation of all three has faced substantive challenges; documented gaps between framework ambition and on-ground operational impact are part of the institutional reality, and Afrimintel's province scoring reflects observed governance and value-chain conditions rather than asserted framework adoption.

Institutional mandate-fit analyses. The platform publishes structural alignment readings against the published strategic frameworks of multilateral DFIs whose mandates are aligned with the platform's coverage. The first published analysis, on the African Development Bank's published strategic priorities (Ten-Year Strategy 2.0, High 5, Natural Resources Management and Investment Action Plan 2025-2029, ANRC framework, Joint MDB Statement on Critical Minerals), is at /methodology/afdb-mandate-fit/. The analysis engages exclusively with publicly-published institutional documents; no individual is named; no relationship is claimed; no commercial engagement is implied. Subsequent multilateral and bilateral DFI mandate-fit analyses follow the same template.

Benchmark Spreads. The platform's discipline of publishing where multiple public sources diverge on the same metric for an asset, the size and source of the divergence, and the most-defensible reading with reasoning. The methodology document at /methodology/benchmark-spreads/ articulates the discipline and the criteria a Benchmark Spread must meet to publish. Spreads on the four published case-study assets (Kabanga Pre-FID, Manono Disputed Tenure, Loulo-Gounkoto Post-Settlement, Lobito Corridor Infrastructure) are nested under each asset's case-study directory and linked from the parent dossier. Roadmap: extend to full intelligence-grade tier across calendar 2026.

Record tier definitions

Records carry a machine-readable grade_class field; a derived tier assignment places each record in one of four tiers, with intelligence-grade defined as the parameter-complete, primary-source-cited subset: 18 intelligence-grade dossiers, with operator-attestation, research-grade, and spatial-reference context tiers defined below.

INTELLIGENCE-GRADE (18 records · grade_class: decision-grade)

Criterion: Backed by a citable source (NI 43-101 Technical Report, JORC-compliant report, audited annual report, or named operator disclosure) with resource statement (tonnage + grade), named operator, and source document identifier.

Intended use: Citable as orientation in a professional context, subject to the reader's own verification against the primary filed document. Afrimintel is not QP-signed.

OPERATOR-ATTESTATION (grade_class: operator-attestation-grade)

Criterion: Resource or production figures sourced from named-operator disclosure (company reporting, investor materials, or operator group statements) but without a QP-signed NI 43-101 / JORC / S-K 1300 framework. Operator and figures are attributed; independent reporting-code verification is not yet complete.

Intended use: Orientation with operator attribution; the figures carry the operator's basis, not Afrimintel's independent verification, and are not a substitute for the primary filed document.

RESEARCH-GRADE (grade_class: research-grade)

Criterion: Named operator and named primary-source citation (national mining survey body, government commerce department, or operator group), but without a published reserve/resource statement against an internationally recognised reporting code. Establishes existence and operator attribution beyond a coordinate, without claiming resource quantification.

Intended use: Orientation only; not citable for capital allocation. Use as the named-asset surface where the underlying jurisdiction does not file under NI 43-101 / JORC / S-K 1300. Includes, among others, six named Rwanda assets — Rutongo, Nyakabingo, Gatumba, Bugarama-Bukunzi, Musanze, NW Rwanda Gold Belt — sourced to the Rwanda Mines, Petroleum and Gas Board (RMB) and the US Department of Commerce Rwanda Mining Guide, alongside other sourced-but-not-yet-parameter-complete records across the platform.

Upgrade path: A research-grade record graduates to intelligence-grade if and when a tonnage + grade statement under an internationally recognised reporting code becomes published.

SPATIAL REFERENCE (no grade_class field)

Criterion: Name, coordinates, commodity, status, province only. No dossier.

Intended use: Map context showing where named deposits exist. Not actionable without primary-source research.

Audit cadence

Afrimintel commits to quarterly review of intelligence-grade records:

• Every 3 months, at least 2 intelligence-grade records (≈10%) are re-audited against primary source
• Records with material errors are corrected immediately
• Records that cannot be re-audited are demoted to spatial-reference tier
• Audit documents are published at /methodology/<record-name>-audit.md
• Each audited record carries last_reviewed and next_audit_due fields; records past their next_audit_due render their INTELLIGENCE badge in amber with OVERDUE suffix

First audit completed: Kamoa-Kakula (18 April 2026). Next audit due 18 July 2026.

Audit source protocol

Every audit MUST compare the Afrimintel record against the PRIMARY FILED DOCUMENT (NI 43-101 PDF on SEDAR+ or filer's website, JORC report, audited annual report). Summaries, press releases, and investor presentations may supplement the audit but NEVER substitute for the primary document.

Rationale: the first Kamoa audit (April 2026) was performed against a summary of the NI 43-101 rather than the filed PDF. This is a structural weakness — summaries drop nuance and can transcribe imprecisely. Future audits must open the primary PDF. The Kamoa record will be re-audited against the primary PDF at the next cycle to close this gap.

Tool reliability testing

Where Afrimintel surfaces a Derived value computed by a tool — currently the DCF / NPV calculator — the tool itself is tested against a published input battery. The battery is published, not just the results, so any user, partner, or external reviewer can run the same tests and verify the tool independently.

The Day 30 v1.0 battery covers 15 tests across 5 categories (monotonicity, boundary, sensitivity, reconciliation, pre-specified failure mode), anchored to the Kamoa-Kakula Copper Complex DRC and its five published economic studies (Kakula DFS 2020 through Kamoa-Kakula 2023 IDP PEA — NPV targets spanning $5.5bn to $20.2bn).

The Q3 2026 expansion of the battery is scheduled for joint authorship with a named external mining-finance reviewer. The v2.0 battery will carry the reviewer's name and credentials alongside Afrimintel's.

→ Read the DCF / NPV Tool Test Battery (v1.0)

Commodity price methodology

LIVE PRICES

Fetched on each daily-brief render from Yahoo Finance:

• Copper — COMEX HG=F
• Gold — COMEX GC=F
• Nickel — CME NI=F

Important: These are COMEX/CME base prices, not LME. Basis differentials may apply.

STATIC PRICES (EST)

Manual refresh at quarterly cadence minimum. Every static price carries a verified_at date shown in the UI:

• Cobalt — LME April 2026 ($56,290/t)
• Lithium — Li₂CO₃ CIF China, Fastmarkets April 2026 ($20,684/t)
• NdPr Oxide — SMM China April 2026 ($108.64/kg)
• Graphite — flake fines estimate April 2026 Derived
• Manganese — CRU estimate April 2026 Derived

Province scoring

Afrimintel Score = (MSP × 0.6 + IC × 0.4) × Opportunity Multiplier

• MSP (Mineral System Potential): 0-10 score based on 5 criteria
• IC (Investment Climate): Normalised Fraser Institute Annual Survey IAI score ÷ 10
• Opportunity Multiplier: 1.05–1.30 editorial adjustment reflecting coverage gap

Current province scores — May 2026 (v1.1)

Country Risk Composite

The Country Risk Composite is a country-level (as distinct from province-level) score combining four named source documents into a single 0-100 risk-attractiveness reading. It is rendered on country profile cards across the platform and is referenced by the AI synthesis layer when a country is named in a query.

Core = (Fraser × 0.40 + TI × 0.30 + RGI × 0.30), re-normalised across available core pillars. EITI = +5 bonus if the country is an EITI implementing member (capped at 100); non-members are not penalised. Suspended members hold membership but earn no bonus — a suspension for inadequate stakeholder engagement negates the disclosure signal the bonus rewards; enhanced-scrutiny members retain it (graded treatment adopted v2.6.0, F-EITI-8 resolution; statuses per the EITI register at the stated retrieval date). (Revised v1.6.3: previously a four-pillar weighted average Fraser 0.30 / TI 0.25 / RGI 0.25 / EITI 0.20; changed to bonus-only so a strong non-EITI jurisdiction is not dragged down by a 20-point structural gap.)

• Fraser (40% of core) — Fraser Institute Annual Survey, Investment Attractiveness Index. 0-100 scale, higher is more attractive. Current source: Fraser Institute Annual Survey 2025, published 26 February 2026. Per v1.1 Pending Fraser PDF back-check discipline (Methodology §07), values for 14 African jurisdictions are surfaced — Seven values are Sourced (confirmed against the Fraser 2025 PDF, 2026-06-01) and feed the weighted composite: South Africa 52.97, DRC 57.46, Namibia 56.21 (direct read of Table 1) plus Botswana 85.99, Morocco 78.97, Ghana 55.21, Zambia 72.84 (PDF-confirmed). The remaining African values still rest on secondary press/aggregators and are Pending PDF back-check — EXCLUDED from the weighted composite, footnoted only: Côte d'Ivoire 60.92, Tanzania 68.04, Guinea 52.16, Mali 46.58, Burkina Faso 35.29, Angola ~54.5, Egypt ~36. Zimbabwe was NOT ranked in Fraser 2025 (insufficient responses) and is Fraser-absent. This tiered model — verified-only in the weighted composite — is deliberate: the sourcing tier is shown transparently rather than weighting unverified numbers.
• TI CPI (30% of core) — Transparency International Corruption Perceptions Index 2025, released 10 February 2026. 0-100 scale, higher is less corrupt. Full Africa coverage in source.
• NRGI RGI (30% of core) — Natural Resource Governance Institute, Resource Governance Index 2021. 0-100 scale, higher is better governance. NRGI has not published an update since 2021; this remains the latest available primary source. Coverage: 17 African jurisdictions.
• EITI status (+5 bonus only) — Extractive Industries Transparency Initiative membership. A current EITI implementing member (Compliant, Candidate, or Suspended) receives a +5 bonus on the re-normalised core (capped at 100); a non-member receives no bonus and no penalty. EITI is not a weighted pillar — this prevents a strong-governance non-member (e.g. Botswana, a top-10 Fraser jurisdiction that has chosen not to join EITI) from being structurally dragged down. Only PDF-Sourced Fraser values feed the composite Fraser pillar (South Africa, DRC, Namibia, Botswana, Morocco, Ghana, Zambia). Pending PDF back-check jurisdictions (Côte d'Ivoire, Tanzania, Guinea, Mali, Burkina Faso, Angola, Egypt), not-surveyed countries, and Zimbabwe (not ranked in 2025) are treated as Fraser-absent in the composite, which re-normalises across the remaining core pillars.

Weight re-normalisation when components are missing

Source coverage is uneven across African jurisdictions. When one or more of the three core pillars (Fraser / TI / RGI) is unavailable for a country, the core re-normalises across whichever pillars are present. The EITI bonus is then applied to the re-normalised core.

Worked example. Senegal has TI (46), RGI (64), EITI (member), but no Fraser 2025 score. Core = (46 × 0.30 + 64 × 0.30) ÷ (0.30 + 0.30) = 33.0 ÷ 0.60 = 55.0; the EITI member bonus adds +5, for a composite of 60.0 (band: LOW). The Fraser weight redistributes proportionally across the remaining core pillars.

Low-confidence flag. When fewer than two of the four signals (Fraser, TI, RGI, EITI-membership) are present, the composite carries a low-confidence flag in the data layer and the AI synthesis layer is instructed to present the score as indicative-only — leading with the underlying Sourced signals rather than the composite number. Currently flagged (one input each): Namibia, Rwanda, Kenya. A visible on-card low-confidence badge is a tracked enhancement.

Risk bands

• LOW — composite ≥ 60
• LOW-MEDIUM — composite ≥ 40 AND Fraser ≥ 65 (the Fraser top-tier override prevents jurisdictions that score highly on investor-attractiveness but happen to be EITI non-members from being penalised into MEDIUM purely on the EITI 20% weight)
• MEDIUM — composite ≥ 40
• HIGH — composite ≥ 25
• VERY HIGH — composite < 25

DRC editorial override

DRC carried a published editorial override (MEDIUM→HIGH) under the pre-v1.6.3 weighted model. Retired at v1.6.6, when DRC computed to HIGH on its then-current inputs, making the override dead code. The subsequent v1.6.8 Fraser correction (the DRC Investment Attractiveness Index, previously mis-entered as the policy sub-index ~18, corrected to the blended IAI 57.46) moved DRC's screening composite to MEDIUM (currently ~45) — accepted by the v1.6.8 ruling, with the explicit caveat that the blended IAI understates DRC policy-perception risk for cost-of-capital purposes; that risk is carried on the rate side by the π_p policy-perception overlay (v1.6.11), not by re-overriding the screening band. This paragraph was corrected at v2.5.0 (removed-mechanism drift) and again at v2.8.2, when a post-deploy check caught the v2.5.0 text still asserting the v1.6.6-era HIGH/~33 outcome as current.

What the composite is not

The composite is a screening tool, not a project-level risk model. It does not capture asset-specific, operator-specific, or contract-specific risk. Country profile cards on the platform display the composite alongside the four input scores so that a reader can inspect what's driving the headline number and adjust judgement against their own deal-specific inputs.

Safeguards layer — gate-and-floors (Component E)

The environmental and social safeguards layer is built as a gate-and-floors architecture — critical-requirement gates, a foundational-management (PS1) cap, per-pillar floors with no cross-pillar averaging, tiered components, context/Absent, and a fully visible decomposition — not a single weighted-average score. It mirrors how the authoritative standards (IRMA, SASB, IFC) actually structure safeguards, and every figure is Sourced, Derived, or Absent. Full architecture, evidence base, and a worked Kabanga example: Safeguards layer — gate-and-floors.

Epistemic tagging convention

• Sourced — confirmed from named citable source
• Derived — directionally likely, derived from related verified data
• Derived — plausible but no direct evidence
• Absent — no reliable basis

Tags are applied at field level. A record tagged "intelligence-grade" may have individual sub-fields at Derived or Absent.

Spatial-to-intelligence upgrade roadmap

Promotion priority:

1. Commodity × country × strategic-interest
2. Availability of primary source
3. Pilot-user-requested records

Near-term upgrade queue: First 20 targets to be published in the next release. Current commitment: 1 upgrade per week from April 2026 onwards.

What Afrimintel is NOT

• Not a substitute for primary-source research (SEDAR+, ASX filings, annual reports)
• Not a real-time market data terminal
• Not an exhaustive database of African mining (Wood Mac, S&P Capital IQ Mines, SNL cover thousands of deposits; Afrimintel covers 18 at intelligence-grade depth, by deliberate editorial choice)
• Not QP-signed. No Qualified Person has reviewed records. Tier labels are the author's judgement against written criteria.
• Not for sale. Beta access by invitation only.

Corrections and feedback

Data errors, corrections, or partnership enquiries: nikesh@afrimintel.com

Every substantive correction is acknowledged and, where appropriate, credited. See the Audit Log for the public record of every correction made since April 2026.

See the methodology applied to a real decision

The Case Studies section publishes worked demonstrations of the methodology applied to live institutional investment decisions. The triptych — Kabanga Nickel Pre-FID DFI Investment Brief, Manono Lithium Disputed-Tenure Capital Allocation Brief, and Loulo-Gounkoto Post-Settlement Re-Entry Brief (9 May 2026) — covers three structurally different decision shapes: pre-FID greenfield screening, disputed-tenure capital allocation, and post-settlement re-entry / risk repricing. Each case study is a self-produced demonstration; no institutional partner has commissioned, reviewed, or endorsed any case study, in keeping with the published Counterparty Extension discipline.

The Track Record page is the platform's published record of corrections caught (recorded in the Audit Log), prediction-resolution items where the platform held a position before a verifiable outcome, and tracked positions currently carried in the data layer awaiting resolution.

To submit a real deal scenario for the platform to be shaped against, see the structured intake at /intake/.

Document version: 1.0 (22 April 2026) · Next review: July 2026 · Afrimintel · Editorial responsibility: Nikesh Patel, Honorary Consul of Rwanda in Mauritius