A research desk behind every ticker you type.
Chatie Agent runs a panel of investor-model agents across live market data, then shows you where they disagree, because the disagreement is the signal, not the headline verdict.
Watch the desk argue
Nineteen mandates, one ticker
Each agent reasons from a fixed philosophy (value, growth, macro, tail-risk) and cannot see the others' output before submitting. You get a spread of views, not one averaged opinion.
The spread is the output
Dissent, itemised
When the panel splits 8-6-5, that is information. Chatie surfaces the exact line item each agent disagreed on, so you can judge whose reasoning you actually buy.
The data the agents read
Every claim in every transcript is linked to a row in one of six underlying datasets. If an agent cites a number, you can click it and see the filing it came from.
Operational KPIs
Sector-specific metrics that drive the thesis
Income statements
Revenue, expenses, profitability (30+ years)
Balance sheets
Assets, liabilities, and equity
Cash flow statements
Operating, investing, financing
Filing excerpts
Section-level 10-K / 10-Q text, verbatim
Insider trades
Executive transactions with timestamps
{
"source": "Delta Air Lines, Inc. (DAL)",
"data": {
"Premium Revenue": {
"value": "$5.36B",
"ref": "+13.9%"
},
"Loyalty Revenue": {
"value": "$1.22B",
"ref": "+12.8%"
},
"Cargo Revenue": {
"value": "$226.0M",
"ref": "+8.7%"
},
"Total Revenue per ASM": {
"value": "22.92¢",
"ref": "+11.6%"
},
"Passenger Yield": {
"value": "21.78¢",
"ref": "+5.6%"
},
"CASM Ex-Fuel": {
"value": "15.13¢",
"ref": "+6.3%"
},
"Fuel Cost per Gallon": {
"value": "$2.78",
"ref": "+12.6%"
},
"Fuel Gallons Consumed": {
"value": "988M",
"ref": "+1.2%"
},
"Total Liquidity": {
"value": "$8.10B",
"ref": "-"
}
}
}Panel telemetry, in the open
Live stats from the last 30 days of panel runs. Not performance claims; the panel has no track record and we won't invent one.
About this telemetry
Dispersion is the gap between the most bullish and most bearish fair-value estimate in a run, as a percent of spot. Wide bars mean the panel genuinely disagrees, which is where reading the transcript pays.
Methodology
- Sample: all public panel runs in the trailing 30 days.
- Each bar aggregates runs by sector, median values shown.
- Raw run transcripts are linked from every data point in the terminal.
Wire the panel into your stack
One endpoint runs the full panel and streams every agent's reasoning back. Build your first request in seconds.
import requests resp = requests.get( "https://api.chatie.agent/panel/run", headers={"X-API-KEY": "<your-api-key>"}, params={ "ticker": "NVDA", "agents": "all", "include": "transcript,sources" } ) run = resp.json() print(run["consensus"]["split"])
What this is not
Worth saying plainly, because most tools in this category won't.
Chatie Agent is a research tool, not an advisor. The agents are language models reasoning over public data. They have no track record, cannot see your portfolio or tax situation, and their fair-value estimates are model output, not forecasts. Nothing here is investment advice. Agents are named after public investors to describe a style of reasoning. None are affiliated with, endorsed by, or speaking for any living person or estate.
Type a ticker. Read the argument.
Five runs on the house. No card, no sales call.