38,207
Companies priced worldwide · ~11,000 US-listed
One harness. Many desks.
Receipts, always.
The question is only the beginning.
Ask: "What changed in this name, what matters for my book, and what should I investigate next?"
Specialist agents.Assembled around the question.
Reads growth, inflation, policy, and rates, and states the regime the trade lives in.
Produces the pre-market brief. Cites the prints and the calendar events that can break the view.
Examines news flow, analyst tone, transcripts, and positioning, and separates the crowd's view from the evidence.
Flags where consensus is thin. Ranks what changed in the name and why.
Runs factor tilts, screens, correlations, and backtests in code, and returns every number with its calculation attached.
Computes portfolio impact. Challenges every other agent's conclusion with the data.
Compares options pricing, skew, and dealer positioning, and structures the trade ticket with entry, invalidation, and sizing.
Tracks unusual volume and flow. Prices the risk in plain English.
Examines estimate revisions, guidance, beat rates, and implied versus realized moves ahead of the print.
Produces the pre-earnings package. Fundamentals, valuation, EPS history, guidance.
BriefHead of Capital Markets, Multi-Asset Desk
I don’t need more information. I need the setup, the invalidation, and the receipts. AgentSmyth delivers that in one place.
PerformanceDWG 002
- Measured by
- AgentSmyth
- Checked by
- You
- Revision
- A · 05 Nov 2026
2,473
US 10-K/10-Q filers, $300M or more
23
Geographies supported
Your environment. Your controls. The same Harness.
Work in AgentSmyth, connect it to your existing systems, or deploy it within your institution's environment. Choose the deployment and model-access controls appropriate to the workflow.
CLOUDCUSTOMER VPCON-PREMAPIMCPCLI
- Question
- How exposed is the book to a 50bp rise in the 10-year?
- Answer
- Down 2.1%, mostly through duration in the bond sleeve.
Receipt 1 of 31
- Cited
- “Down 2.1%”
- Source
- FRED
- Document
- 10-year Treasury yield, daily
- As of
- 04 Nov 2026
- ID
- DGS10
What produced this answer has an answer.
Real desks. Real decisions.
Every run, on the record.
The Harness in action
Build the function your terminal never shipped.
- Ask
- Rank the names in my book reporting in the next 30 days by implied move against their last eight realized moves.
- Card
- The run is saved as a card: the data calls it made, exactly as they ran, and the script that does the ranking.
- Desk
- Pinned to your Semis desk and run every weekday at 06:45 ET on fresh data. Your team gets the method, not a screenshot.
Pre-earnings checkSemis desk
Names in the book reporting in the next 30 days
NameImplied move vs. last 8ImpliedLast 8Gap
MU±9.4%±7.1%+2.3
MRVL±10.8%±9.6%+1.2
AVGO±6.9%±6.1%+0.8
NVDA±7.2%±6.8%+0.4
AMD±8.1%±8.6%−0.5
Run · Mon · 06:45 ETRanked by gap
1Plan
- 01Holdings · Core Book
- 02Report dates · 30 days
- 03Implied moves · options
- 04Realized moves · last 8 prints
2Script
gap = implied - abs(last_8).mean()Re-run at save · output matches3ScheduleWeekdays · 06:45 ETNext run —
4DeskSemis deskShared with your team
CEO's Weekly Drop
Shipped this month
- Portfolio and watchlist changes, with approval
- Cards that act
- Desks shared across the team
Specialists do the work. Evidence stays attached. You approve what changes.
See and shape the system behind the work.