HO
Use cases / Research
Home Bot
A six-layer research terminal at home. Data, news, alerts, the dashboard.
Author names, descriptions and prompts are preserved as published. Need a starting point? Read the bot selection guide →
The story behind this use case
Roan (backend, HFT-style execution, quant systems) starts honest: nobody replicates Bloomberg's licensed data moat at home. Real-time exchange feeds are contractually locked. A terminal is seven layers — Data, Analyst, News, Reasoning, Report, Alert, Visual Dashboard — and only the Data layer is protected by contracts. The other six are workflow and integration, and that's the part that just became solvable at consumer scale.
His numbers: a Bloomberg seat is about $27,660 a year; this setup is roughly $3,000 a year for about 70% of the utility. Prior walls he'd hit for a year: real-time monitoring across hundreds of sources, and dashboards that needed a frontend engineer.
This is a build guide, not a P&L claim. Analysis only, no orders.
Screenshots & demos
Prompt
This reference prompt was supplied by its author. Review it before giving it to your agent.
Build me a seven-layer market workstation on this computer. Seats: RETRIEVAL (pull prices, filings, economic data on schedule), ANALYST (track ratings changes and where consensus diverges from price), NEWS (monitor my source list continuously, dedupe, timestamp), REASONING (answer my questions against only what's in the data folder), ALERTS (fire on my defined price/news/fundamental conditions), REPORTS (morning macro note, earnings previews, position summary), DASHBOARD (render a live local page: charts, heatmap, blotter, correlation matrix). Never state a number without its source and timestamp. No orders, no spending — analysis only.