stagirus

We measure the distance between what was announced and what has been confirmed, using primary records and their conditions.

01

AI

We open the paper, the model card, and the evaluation and reproduction record together, and state what a new model or study has shown - and what it has not yet shown.

02

Hardware

Accelerators, fabrication, memory, networking, datacenters and power constraints, read against supply schedules, performance conditions and cost structure to locate the real bottleneck.

03

Products

The current state of products and tools by version, access conditions, real working sessions and failure modes - keeping demos separate from shipped capability.

04

Policy

Bills versus final rules, jurisdiction and effective dates - and corporate announcements read against filings and signed contracts to see what obligations actually changed.

Fact Check

Check an official announcement's central claims against the primary record, separating announced, demonstrated, limited availability, general availability and independently confirmed - and state what can be verified today.

The official announcement plus, where they exist, the paper, model card, filing or rule text; version, date, availability scope, and what remains unconfirmed.

Explainer

Unpack a complex paper, model card, rule or technical document into short verifiable units - claims and limits - so the reader can return to the original.

At least one primary document with exact version and date, the key limitations, the location each claim maps to, and a link the reader can check directly.

Analysis

Read performance numbers inside their task, data, baselines, runtime environment and reproduction status; results from different conditions are never ranked on one axis.

Task, dataset, metric, baselines, hardware or inference settings, versions, and whether results are self-reported or independently reproduced.

Review

Record where a product or tool works in a concrete real task, what access it required, and what failed repeatedly - claiming no standing test lab beyond the article itself.

Per-article method, exact version, access tier, duration and scenario, observed failures, comparison scope, and the limits of what that method could confirm.

Report

Read corporate announcements against legal filings and disclosed contractual obligations, marking proposed, signed, closed, conditional and executive outlook as different states.

The primary filing or contract record, the parties, signature or proposal status, disclosed amounts and terms, and material undisclosed conditions.

Masthead

Bylines and the verification process behind every article are described in our editorial policy.

Products & Policy Desk

Daniel Koo

Splits time between what a product actually does in a working session and what a policy or corporate announcement actually obligates. A feature exists when it runs outside a demo; a rule takes effect on the date the text says it does, not the date of the press release.

Products & ToolsPolicy & Filings

AI Research Desk

Elliot Marsh

Covers new models and studies by reading the paper, the model card, and the reproduction attempts side by side, then reporting only what the evaluation record actually supports. A claim of a new capability gets checked against the benchmark's conditions before it gets repeated as fact.

AI ResearchBenchmarks

Compute & Infrastructure Desk

Priya Raghavan

Tracks accelerators, fabrication, memory, and networking against actual supply schedules and delivery dates, not roadmap slides. The question on this desk is always where the real constraint sits: wafer capacity, packaging, power, or contract terms, and whether an announced timeline matches what customers were actually promised.

ChipsDatacentersSupply

stagirus