But it kept a little less of each dollar last time.
BizBoxAI trading desk
Quorum
AI agents. Research. Debate. Trade.
Watch their own practice desk at work.
Practice money · not financial advice
Watch the room work · sample company
Q
Northstar Co. · sample company
Your desk read 14 sources and found five things worth checking.
Some are excited, some are unsure.
The source and time are saved for you.
If the delay is short, things look good.
If that customer leaves, sales drop fast.
Two answers are still missing. We tell you which.
Research, not advice. The call is yours.
QUORUM · LIVE DASHBOARD
Total$—
Cash$—
Invested$—
AAPLHold
NVDABuy
Scroll to enter the live dashboard
Several minds look at the evidence first. Then you decide, with the sources in front of you.
- Sources shown, with times
- A human makes every call
- Not financial advice
01 / Who it’s for
From your first market question to a seasoned research process, better decisions start with seeing more than one side.
Learning the market
Build confidence with a clear explanation of the business, the opposing case, and what you still need to learn.
Research on your terms
Look beyond a price chart. Explore the evidence, compare viewpoints, and keep risk visible as you follow a company.
Pressure-test the thesis
Review assumptions, counterarguments, and uncertainty in a structured brief, with final judgment firmly in your hands.
02 / A better research rhythm
No single signal gets the final word. Each perspective has a job, then the evidence comes together in one clear brief.
- 01Analysts
Four researchers each read a different kind of information: the money (), the mood (), the news and the price history.
- 02Bull & bear
One argues the hopeful side (the ), one argues the worried side (the ), using the same facts.
- 03Research lead
Listens to both and writes down what has to be true for each side to be right.
- 04Risk team
Points out what could go wrong (the ) and what we still don’t know.
- 05You
You read a short summary, check it, and make your own choice. Quorum never buys or sells for you.
03 / Try it
Tap a question and watch Quorum work it out loud: who says what, which way the room leans, and what it’s still unsure about. A made-up sample company; no real data.
Ask Quorum about Northstar Co. · sample company
Research lead
Fundamentals
Sentiment
News
Technicals
Bull
Bear
Risk team
04 / What you walk away with
Every question ends here: one paragraph, both sides, how sure the desk is, what to check, and where every fact came from. Switch the level at the top to see it change.
Research brief · Northstar Co.Sample company · Oct 5, 2026 · 9:41 AM
Research, not adviceNorthstar is selling more each quarter, mostly thanks to a new product. Two things could change the picture: it is keeping less of each dollar, and one big customer buys a lot. We are medium-sure, and here is what to check.
- Sales up three times in a row
- The new product is selling
- Handled delays before
- Keeping less of each dollar
- One customer buys a lot
- A supplier is running late
LowMediumHigh
Medium: two important answers are still missing (below).
- Quarterly reportCompany filing · sample
- Earnings callTranscript · sample
- Supplier updateNews article · sample
- Social conversation1,200 posts · sample
05 / Watch it trade
The desk runs its own $10,000 paper account. Every weekday after the market closes it researches its watchlist, makes its calls, and trades at real market prices, with the reason for every trade. No real money, and never yours.
Simulated $10,000 · Paper trading · not real money · not investment advice
Started$10,000.00
Total value$9,676.12
Cash$8,087.96
Invested$1,588.16
Paper trading · not real money · not financial advice
AAPL9.9%
GOOGL5.0%
MSFT1.2%
AMZN0.3%
Cash83.6%
| Ticker | Qty | Price | Value | Wt |
|---|---|---|---|---|
| AAPL | 2.82 | $339.11 | $955.95 | 9.9% |
| GOOGL | 1.74 | $279.03 | $484.73 | 5.0% |
| MSFT | 0.23 | $507.13 | $117.91 | 1.2% |
| AMZN | 0.12 | $245.80 | $29.57 | 0.3% |
| Date and time | Symbol | Side | Quantity | Price | P&L |
|---|---|---|---|---|---|
| GOOGLPractice money only · simulated · not real money · not financial advice | Sold | 1.7372 | $279.03 | -$1.27realized | |
| NVDAPractice money only · simulated · not real money · not financial advice | Sold | 1.1231 | $202.20 | -$28.66realized | |
| MSFTPractice money only · simulated · not real money · not financial advice | Sold | 0.2326 | $507.13 | -$4.73realized | |
| AAPLPractice money only · simulated · not real money · not financial advice | Sold | 2.819 | $339.11 | -$17.68realized | |
| GOOGLPractice money only · simulated · not real money · not financial advice | Bought | 3.4744 | $279.76 | -$2.54open | |
| NVDAPractice money only · simulated · not real money · not financial advice | Sold | 1.1232 | $204.81 | -$25.73realized | |
| MSFTPractice money only · simulated · not real money · not financial advice | Sold | 0.4651 | $521.90 | -$2.60realized | |
| AAPLPractice money only · simulated · not real money · not financial advice | Bought | 2.8041 | $344.80 | -$15.96open | |
| GOOGLPractice money only · simulated · not real money · not financial advice | Sold | 1.7122 | $283.75 | -$68.15realized | |
| NVDAPractice money only · simulated · not real money · not financial advice | Sold | 2.2463 | $213.72 | -$31.44realized |
Showing 10 of 45
Paper money, simulated, not investment advice. A sale’s P&L is against the average cost of the shares sold; a buy’s is how far the latest price has moved since.
Underweight · whyReduce GOOGL exposure immediately to align with the confirmed downtrend while guarding against the asymmetric snapback risk of an extreme oversold condition on the eve of a binary political event.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Sell · whyThe agents rated NVDA Sell. Their note quoted price levels, which this page leaves out.
The rule · exitSell: our rule sells the whole position. It never shorts.
Underweight · whyRating: Underweight Executive Summary Hold the existing 0.4651 units of MSFT; do not trim, do not add new capital at spot. The research manager’s strategic plan explicitly instructs that if the position is already below 0.5× a standard full allocation unit, the trim should be skipped. The current holding sits below that threshold, so the sale proposed by the trader is halted. The Underweight rating stands.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyUnderweight AAPL. Execute a partial trim of the existing 5.638‑unit position to bring it to an underweight allocation relative to the portfolio benchmark, targeting 50–75% of the normal stock weight. If the current weight equals the normal allocation, this implies selling roughly 25–50% of the stake (approximately 1.4–2.8 units); otherwise, scale the sale to meet the underweight target.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Overweight · whyDo not add the full proposed 0.65 shares at once. Instead, wait for a confirmation signal before deploying the second tranche.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Underweight · whyExecute the trim only under one of two data‑confirmed scenarios.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyThe agents rated MSFT Underweight. Their note quoted price levels, which this page leaves out.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Overweight · why- Rating: Overweight - Executive Summary Maintain the existing AAPL core holding of 2.834 units; do not trim. Overweight remains the correct stance, but the near‑term binary‑event risk (earnings on November 2 and the midterm elections on November 3) makes immediate new buying imprudent.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Sell · whySell the entire 1.712-unit GOOGL position immediately. The Aggressive Analyst’s regime-change thesis decisively outweighs the Conservative’s capitulation argument and the Neutral’s compromise.
The rule · exitSell: our rule sells the whole position. It never shorts.
Underweight · whyUnderweight NVDA. Place a conditional re‑entry order.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyUnderweight AAPL.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyRetain the remaining 2.285 units but replace the initial hard stop with a tiered exit rule.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyRating: Underweight Executive Summary Execute the trader’s proposal to sell 0.12 units of AMZN at the prevailing price (reference 254.60), reducing the position to approximately 0.12 units. This locks in partial profits on the existing long and lowers portfolio beta ahead of the October 29 earnings event. Place a hard stop on the remaining position.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyReduce NVDA exposure by approximately one-third to one-half in stages.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Overweight · whyOverweight GOOGL, but do not execute the full trader-proposed buy before the October 28 Q3 earnings event. Keep the current 0.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Underweight · whyUnderweight AMZN. Do not add new capital.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Overweight · whyOverweight NVDA and begin building the planned overweight position, but with a reduced initial tranche rather than the full 4 units.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Underweight · whyThe agents rated MSFT Underweight. Their note quoted price levels, which this page leaves out.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Buy · whyRating.
The rule · up to 20%Buy: our rule takes the position up to 20% of the book.
Underweight · whyUnderweight GOOGL.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyThe agents rated AMZN Underweight. Their note quoted price levels, which this page leaves out.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyReduce NVDA exposure without abandoning the long-term position.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyThe risk-analyst debate confirms the Research Manager’s Underweight call and the trader’s proposal to reduce exposure, but the right size lies between the aggressive 50% trim and the conservative 30% hold.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyUnderweight AAPL ahead of the Nov 2 earnings event. This stop is a disciplined reference for normal trading, but it will not protect against an earnings gap.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyUnderweight AAPL; make a measured trim of roughly 0.30–0.59 of the 2.957 units held (10–20%), subject to verifying the current quote, trading constraints, and the portfolio's actual AAPL allocation.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyUnderweight GOOGL: after checking a current quote and execution terms, sell approximately 0.356 of the existing 1.424 units (25%), retaining roughly 1.068 units; this is a partial trim, not an exit or a claim that the remaining position equals 75% of a defined standard allocation.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyUnderweight AMZN: retain a core position but trim the current 1.924 units in stages only if a defined portfolio risk limit shows the position is oversized; otherwise, do not add on this pullback. The proposed 0.75× standard allocation cannot be converted to a sale quantity because the standard allocation and portfolio-level exposure limits are unspecified.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Overweight · whyDo not automatically build to 1.25 times standard.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Underweight · whyUnderweight AAPL. Confirm the standard allocation and portfolio weights before sizing any further sales toward the research plan’s roughly 75%-of-standard target.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyUnderweight AMZN, not an outright exit: retain exposure but trim the 3.848-unit holding only if a verified portfolio-weight calculation shows it exceeds roughly 75% of the firm's defined standard position. The standard position and other holdings' market values are missing, so no sale quantity can responsibly be specified; until the limit is established, make no trade or addition.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Underweight · whyKeep GOOGL as a smaller core holding. Review current market values and the portfolio’s standard single-name and technology-sector limits, then trim the 2.849-unit position in stages only if it exceeds roughly 0.5–0.75× the standard allocation; otherwise make no trade. No exact sale quantity or mechanical stop is justified by the missing allocation data.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Overweight · whyRefresh the quote, trend and volatility first.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Overweight · whyCheck portfolio concentration and size the tranche for tolerable loss, including possible stop slippage.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Underweight · whyUnderweight GOOGL: trim approximately 30% of the existing 5.698 units (about 1.71 units), retaining roughly 3.99 units rather than exiting. Size the remaining holding against the portfolio's actual standard allocation, which has not been provided. Review over the next one to two reporting periods.
The rule · halveUnderweight: our rule sells half the position. A size the agents mention is not used; the rule sizes every order.
Overweight · whyConfirm the current quote and portfolio weights first; if MSFT is below the intended limit, add in up to two roughly 0.10x-standard-allocation tranches toward, but not beyond, 1.20x standard allocation, with the first on a refreshed pullback rather than by chasing strength.
The rule · trim to 20% capOverweight adds up to 10% of the book, but never past 20% in one company. This position had grown past 20%, so the rule sold the part above the cap.
Overweight · whyOverweight AAPL modestly, but do not buy at the historical trigger without checking live prices. Stage any further additions toward, but not automatically to, 1.1–1.2× a standard allocation.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Overweight · whyOverweight GOOGL through a small, conditional addition rather than an immediate market buy.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Overweight · whyFavor a measured increase in NVDA, not an immediate purchase. Treat 0.75x a standard allocation as a provisional ceiling, not a mandate.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Overweight · whyBefore ordering, set a portfolio-level exposure cap and loss budget accounting for MSFT, NVDA, and GOOGL.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Overweight · whyThe agents rated GOOGL Overweight. Their note quoted price levels, which this page leaves out.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Overweight · whyOverweight AAPL through a conditional, small starter purchase, not an immediate full allocation.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Overweight · whyCap total exposure at 50%–75% of a standard allocation only after defining portfolio and single-name risk limits, and do not add if already above them.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Overweight · whyFavor a measured increase in MSFT, not an immediate full-size order: verify the live quote, portfolio weights and allocation baseline before adding, and make a small first addition only if MSFT remains below the intended allocation and a pullback stabilizes or strength is reconfirmed. Stage any further increase toward at most roughly 1.25 times a standard allocation.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Overweight · whyOverweight NVDA through a cautious initial purchase, not an immediate 1.25x-standard allocation.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Overweight · whyOverweight MSFT modestly.
The rule · add up to 10%Overweight: our rule adds up to 10% of the book, never past 20% in one company.
Date2026-11-02UpdatedNov 2, 10:29 PM UTCDecisions5 tickers
Fixed rules size every trade; the AI never picks a size. Buy = up to 20% of the book. Overweight = add up to 10%, never past 20%. Underweight = sell half. Sell = exit, never short. Prices are real market prices.
06 / Prediction markets
Watch a paper bot quote one cent inside a Kalshi demo spread. It only records simulated fills when a later book crosses a prior quote.
Practice money only · simulated · not real money · not financial advice
No Kalshi demo feed is running yet. The stock paper desk above remains available.
Under the hood
TradingAgents
TauricResearch’s multi-agent research framework, running on our server. Apache-2.0.
View repository ↗Quorum desk.py
Our fixed-rule sizing and SQLite trade record. Alpaca PAPER mirror is limited to the paper endpoint.
Kalshi DEMO
Read-only demo order book, local paper quotes and simulated fills. No exchange orders.
07 / Built with boundaries
01
Show the sources.
Live research should identify where each claim came from and when it was retrieved.
02
Keep real money out.
The agents can record trades in their own practice account. They cannot place real orders or manage member money.
03
Say what is unknown.
Conflicting evidence and missing data belong in the brief, not beneath the interface.
Questions
No. Quorum never trades for you, never touches your money, and never connects to your brokerage. Every decision is yours. The only trades it makes are in its own pretend $10,000 paper account, shown on this page so you can watch how it thinks.
No. It is a simulated $10,000 that only exists on our server. Trades fill at the real market price at that moment, so the decisions are real, but no money is ever bought or sold.
No. Quorum gives you research to read and questions to check. It doesn’t know your situation, and it doesn’t tell you what to do with your money.
No. Switch the page to "New" and everything is explained in plain words. Tap any underlined word to see what it means.
A live version would read a company’s official reports, earnings calls, news and public conversation, and show you each source with the time it was read. This page uses a made-up sample company.
One opinion can miss things. Quorum has researchers argue opposite sides on purpose, so the weak spots show up before you decide.
Quorum is still a research preview. Pricing isn’t set, and there’s nothing to buy on this page.
From BizBox
A research workspace that feels as considered as the decisions it supports.
Ask the roomResearch preview. Your email only puts you on the list; no billing is active, and the only trading is the desk's paper account.