Methodology and disclosures

Coverage focuses on AI infrastructure, model economics, agentic software, and the companies and tools shaping those systems. Articles distinguish reported facts, calculations, interpretation, and forward-looking scenarios.

Primary sources are preferred: company filings, earnings materials, official product documentation, model cards, research papers, benchmark providers, government releases, and direct statements.

Secondary reporting is used for context and triangulation. Material claims should be linked where a stable public source exists.

Calculations are based on disclosed assumptions. Model comparisons consider more than headline token prices, including token consumption, retries, latency, task completion, context handling, and supervision where data is available.

Benchmarks are treated as evidence, not proof of broad production performance.

Forward-looking pieces aim to state the central thesis, the evidence supporting it, and the developments that would weaken or invalidate it. Unknowns should remain visible rather than being converted into false precision.

AI tools may assist with research organisation, transcription, calculations, drafting, editing, and code. They are not treated as authoritative sources. Material claims are checked against cited evidence before publication, and final editorial responsibility remains with Upcurious.

Upcurious content is educational and does not constitute investment, legal, tax, or financial advice. Coverage of a company or asset is not a recommendation to buy or sell it.

Material commercial relationships or specific conflicts relevant to an article should be disclosed within that article.

Substantive factual errors are corrected promptly. Material changes should update the visible modification date and, when useful, include a correction note. Formatting-only changes do not alter the original publication date.

Send correction requests with supporting evidence to theupcurious@gmail.com.