Nyquist
CLOSED BETA · 2026
Start a 60-day pilot →
Nyquist
Every institution runs its risk-and-research on tools nobody designed for AI. The incumbents are priced for the top and assume humans orchestrate. What a PM under $500M AUM gets instead: a ticker in at the close and a debated call by the open, the whole book repriced overnight, and every number opening to the filing and the tick behind it — built for agents, audit-grade, priced for the mid-market. That gap is the opening.
An LLM bolted on top adds a seam, doesn't close one. Every new question becomes another script someone has to maintain.
No vendor ships suptech with audit trails procurement accepts. Every supervisor hits the same wall.
Hebbia and Rogo do document Q&A, not the full lifecycle. Everyone else self-builds and pays the salary bill forever.
Drift quarterly. No audit trail. Brittle handoffs at every seam. Each red marker is a place data gets dropped on the floor.
Tool-use and reliable reasoning have reached the point where an agent can run a stress test and defend the number — not just summarise a document.
Boards now ask how AI is used in risk. The demand is top-down — but the audit-grade, on-prem product they can actually buy doesn't exist yet.
Regulated buyers can't ship a black box. Reproducibility and lineage are table stakes — which is exactly where general-purpose AI tools fall down.
Nyquist is the agent layer that uses them as inputs — full investment lifecycle, full supervisor toolkit, one engine. Underneath: a typed financial ontology with end-to-end lineage from filing to number.
| Capability | Terminal data | Incumbent risk platforms | Doc-AI / Q&A tools | Network analytics | Nyquist |
|---|---|---|---|---|---|
| Full quant primitives | via 3rd party | ✓ | ✗ | partial | ✓ |
| Document research + AI Q&A | partial | partial | ✓ | partial | ✓ |
| Multi-agent swarm | ✗ | roadmap | partial | ✗ | ✓ |
| MCP server · agent-callable | ✗ | ✗ | partial | ✗ | ✓ |
| Own domain SLM | partial | ✗ | ✗ | ✗ | planned |
| On-prem deployment | partial | ✗ | ✗ | ✓ | ✓ |
| Multi-layer contagion | ✗ | partial | ✗ | partial | ✓ |
| Regulator workflow | ✗ | ✗ | ✗ | partial | ✓ |
| Typed ontology + lineage | ✗ | partial | ✗ | partial | ✓ |
| Built for funds under $500M AUM | ✗ | ✗ | ✓ | ✗ | ✓ |
Terminals are news + data + chat. Incumbent risk platforms assume humans orchestrate. Doc-AI tools do document Q&A. Network-analytics vendors do AML graphs.
Nyquist is the agent layer that uses all of them as inputs — full investment lifecycle, full supervisor toolkit, one engine.
A narrow wedge. The whole industry underneath.
The people who validated bank IRB models and built supervisory tooling are now building the product they wished existed — which is why the audit trail and the on-prem posture were there from line one, not retrofitted.
★ Full founder & team detail on the Team page →
We talk to a small number of investors who understand institutional finance and infrastructure. The fastest path is a direct intro.