Nyquist
CLOSED BETA · 2026
Start a 60-day pilot →
Nyquist
For the PM who is also the risk desk. A swarm of agents reads the filings, argues the bull and the bear case, and hands you a call with the reasoning shown. Open any step and check the work yourself — it's not a chatbot or a black box.
The buy-side stack is glue — Jupyter for research, Excel for the last mile, Bloomberg for data, Python for risk. None of it was built for agents; an LLM bolted on top adds a seam instead of closing one. And you can't out-hire it: the fund across the table has twenty analysts on the same names you cover alone.
Jupyter, Excel, Bloomberg, Python — five tools by hand across one pipeline, none designed for agents.
Twenty analysts on the other side, on the same names. Adding headcount at your AUM doesn't pencil out.
Aladdin and Barra run $500K–$2M/yr with six-month installs. Hebbia and Rogo stop at document Q&A.
Legacy stack: pricing from one vendor, risk from another, OMS somewhere else, reconciliation at 7am. Nyquist puts it on one data model, one compute layer, one audit log — every new primitive compounds.
Pricing, risk, stress, scenarios and trading on a single typed model — end-to-end lineage on every value.
One typed model →Options, bonds, swaps and vol surfaces; VaR, Greeks and scenario P&L — all on one engine, no exports.
The engine →Every primitive compounds — the rest of the stack can use it the day it ships, with audit on every call.
Replayable →Reprice the whole book overnight and read what moved, what broke and which hedge got expensive — on the same numbers the agents argue over. Any figure opens to the filing and the tick that produced it.
Each agent runs on a different investor's playbook and scores the trade on its own. They disagree on purpose. You get the synthesis — a confidence-weighted call with an explicit "here's what would make us wrong," and every step open to inspect.
The full demo runs in your browser — no calls, no scheduling. A pre-loaded $120.4M book (21 positions, 4 asset classes) is wired in. Watch the agents debate, run stress, and inspect every reasoning step.