Mineralis — AI-native Equity Research for Mining Equities
Mineralis is a SaaS platform for investment research in the global mining and energy market: every listed mining company in the world gets a self-updating AI research profile with source attribution, material events are detected and classified in real time, and an explainable scoring framework (MAS-Score) replaces black-box DCF. Positioned as 'Bloomberg for junior miners' — the depth and coverage of CapIQ/S&P (USD 30K/year) with AI-first architecture and transparent pricing.
What Mineralis solves — the coverage/speed/trust triad
Investment research for mining equities is a classic vendor lock-in game: S&P CapIQ and Bloomberg charge USD 30,000+ per year and cover junior miners poorly. Specialized tools (MineGPT, VRIFY) are either reactive (Q&A on uploaded reports only) or sell-side-focused (IR tool for the companies themselves, not for the investor). Nobody solves all three problems at once: full coverage of ~5,000 global mining/energy companies including junior miners, real-time detection of material events within minutes instead of days, and source attribution for every claim (AI hallucinations are deadly in investment decisions). Mineralis is exactly that coverage/speed/trust triad.
MAS-Score — explainable valuation instead of black-box DCF
Instead of a DCF valuation with twenty assumptions, Mineralis delivers a 0-100 score per company built from eight expandable sub-scores: Geology (ore quality, 25 percent), Management (track record, 15 percent), Jurisdiction (country risk, 15 percent), Capital Structure (cash, burn rate, 15 percent), Stage (exploration to production, 10 percent), Permitting (10 percent), Catalysts (5 percent), and Promotional Risk (5 percent negative, pump-and-dump indicators). Each sub-score is AI-explained with concrete source attribution — page number in the NI 43-101, insider filing link, USGS comparative data. Score history shows evolution over time; material events trigger automatic recomputation. The Promotional Risk sub-score addresses a real mining problem: pump-and-dump patterns on junior stocks. Indicators: PR-volume-to-substance ratio, insider sales following positive news, inferred-heavy resources without drilling program, historical CEO track records.
Where we stand today and practice proof
Status May 2026: foundation complete (plan v1 tasks 1-12 plus 15-17 of 22), vertical slice for the first company (Lundin Gold) — data pipeline for SEDAR+/EDGAR/ASX runs, PDF parser is TypeScript-native (unpdf, with page-boundary preservation), chunker plus Voyage-3 embeddings index into Qdrant, hybrid search combines semantic and structured hits. Live-service-dependent tasks (Stripe pricing, AI Gateway production routing) are deferred until service accounts are provisioned. P1 roadmap (mining core, ~500 companies, 8-10 weeks) is in execution; P2 oil and P3 gas follow gate-triggered after ARR threshold or clear user feedback. Stack is Next.js 16 plus Clerk, Drizzle ORM on Neon Postgres (aws-eu-central-1) with row-level security via tenant_id, Qdrant Cloud (eu-central-1) for embeddings, Vercel Functions on fra1 — meaning: server location Europe, Swiss-nDSG-compliant build, Stripe USD for a global investor audience. Mineralis matters to our advisory because it shows how a complete AI-native vertical SaaS — coverage pipeline, real-time event detection, source attribution, own scoring framework — gets from spec to productive foundation in a focused sprint. We bring these architecture patterns into mandates where customers plan their own vertical AI products.