Worldview

Generate macro synthesis analysis from quotes in the database. The synthesis will analyze trends by category and cite specific quotes that support each claim.

Track tailwinds and headwinds across industries based on sentiment analysis of earnings call quotes.

Government economic indicators from FRED (Federal Reserve Economic Data). Data updates automatically when new releases are available.

View all companies organized by industry classification.

Live odds from Polymarket and Kalshi for markets relevant to the investable universe. Prices represent implied probability (e.g. $0.72 = 72% chance).

Token consumption and estimated cost across every Anthropic, OpenAI, and Gemini call this app makes. Logged automatically by SDK instrumentation; caller is derived via stack inspection (module + function).

Aggregate token usage across all OpenRouter users. Shows which models and apps are seeing the most demand. Data from openrouter.ai/rankings.

Tracks GPU instance pricing across 5 providers (vast.ai, AWS, Azure, GCP, CoreWeave) for both spot and on-demand. Snapshots taken 3x daily. The spot-to-on-demand spread is a real-time GPU demand signal.

Tracks 16-vCPU general-purpose and compute-optimized instance pricing across AWS, Azure, and GCP — both x86 (Intel/AMD) and ARM (Graviton/Cobalt/Axion), spot and on-demand. Snapshots taken 3x daily. Spot pricing and ARM utilization are leading indicators of agentic-workload CPU demand.

Daily NPM download counts for @anthropic-ai/claude-code and @openai/codex, a real-time proxy for relative adoption of the two coding-assistant CLIs. Source: public NPM downloads API. Today's data lags by ~24h.

Per-company SDK download meters (Twilio, Datadog, Amplitude, Elastic, MongoDB, Snowflake, Bandwidth, Okta + sector trackers) paired with SEC-filed revenue overlays: a computed downloads→revenue verdict scoreboard, the prediction desk, and a forward-quarter revenue prediction under each overlay derived from its backtest. The ecosystem-wide view lives in Data Science → Developer SDKs. Lighter bars = report-seeded months; solid = live API data accumulating in our DB daily.

The Arete "Token Drift" recreation: ecosystem-wide GenAI SDK downloads from npm, PyPI and Hugging Face (+ Docker/Homebrew extensions) — demand, share and stack-structure lenses. Lighter bars = report-seeded months; solid = live API data accumulating in our DB daily. Single-name company meters live in the Infra Software Agent tab; data refresh is shared with it (daily scheduler or its Fetch Latest button).

Tracks DRAM and NAND flash spot/contract pricing from TrendForce (DRAMExchange). Covers DDR3/4/5 chips, GDDR, modules, NAND flash, TLC wafers, and SSDs. Scraped daily.

Bottom-up wafer/exabyte supply-demand model for DRAM and NAND (HBM tracked as a DRAM segment). Headline is the sufficiency ratio = production ÷ demand − 1 (negative = undersupply). Seeded from Bernstein S&D models and maintained daily from worldview ingestion + external research; material driver changes wait for review.

What 1GW of accelerator compute costs, how much of it is memory at current prices, and how memory price moves flow into token costs — then a demand-elasticity sensitivity applied back onto the Memory Agent supply/demand build. Sliders are display-time only; the memory model itself is untouched.

Head-to-head tests of frontier and open-weight models on real worldview agentic jobs: identical prompts, tools, and frozen inputs — only the model changes. Composite quality scores (1–10) with per-dimension detail; full methodology and scorecards expandable below each chart. Round 2 reruns the program on fresh same-session fixtures with standing KB facts in the memory evidence (production-faithful) and Kimi K3 added across all three benchmarks.

TSMC monthly consolidated revenue from SEC EDGAR 6-K filings, with quarter-to-date tracking vs consensus. SIA global semiconductor sales with regional breakdowns.

Scans every public commit on GitHub for AI co-author signatures (Claude Code, Copilot, Cursor, Aider, Devin, Codex, Cline, Windsurf) via GH Archive / BigQuery.

Historical: full commit message scan via BigQuery (Mar-Oct 2025). Recent: BigQuery for totals + GitHub Search API for AI tool counts (Oct 2025+).

Daily data feeds tracking AI infrastructure economics: GPU spot pricing, open-source activity, model ecosystem trends, and data center energy demand.