The deterministic foundation your AI can sit on
Capital markets firms are putting AI — Claude, Copilot and internal models — in the hands of analysts, portfolio managers, risk teams and traders. AI can see more, move faster and uncover possibilities. But it is probabilistic, and pricing and risk can't be. Without a rigorous engine underneath, an AI agent produces a number no one can reproduce, defend to a regulator or reconcile across desks.
NxCore Analytics Services (NAS) closes that gap. The AI does the language; NAS does the math. NAS is Numerix's cloud-hosted calculation engine, in production for six years and exposed through an API written in the language of finance. It returns the same number for the same inputs, on every desk, from any AI tool, and it can show you how it got there.
The AI asks. NAS answers.
NAS speaks the language of trading and risk — instruments, curves, sensitivities and Greeks — rather than low-level developer payloads. A call reads FixFloatSwap(notional, floating_rate_index, maturity, fixed_coupon), so a quant, a PM and an LLM all understand it, LLMs write correct calls against it, and people stay in the driving seat.
Agent identity, audit trails and permission scoping are built in, so every AI-driven calculation is controlled, logged and reproducible. Customization happens at the AI layer; control lives in NAS.
Three integration paths
Python API
Interactive sessions for sub-second pricing and risk. NxCore Python maps almost one-to-one, so existing code carries over.
REST API
Integrate into existing systems with an OpenAPI schema, continuous deployment and a backward-compatibility guarantee.
MCP server
Claude, Copilot, Cursor and other agentic platforms call NAS in natural language and get deterministic results back. Valuation tools at launch, in beta; broader coverage on the roadmap.
Built for the whole firm, from CTO to trader
One engine and one endpoint serve the leaders accountable for the number and the teams who use analytics every day.
For portfolio managers, traders and quant developers
Get pricing, risk and scenario analysis in the flow of work — and build on it fast.
Self-serve through an interactive Python API, a REST API or natural language, instead of queuing behind scarce specialists.
Use NAS to:
- Run sub-second pre-trade pricing, risk and scenario analysis in interactive sessions
- Price instruments in natural language through the MCP server
- Build pricing and risk apps quickly with a financial-language API that LLMs write correct calls against
- Go from idea to working app in hours, not sprints, with or without financial-engineering expertise
- Check the AI's work: reproduce any result as a notebook via the diagnostics endpoint
- Bring your own market data, or use built-in reference and market data for supported instruments.
For technology and risk leaders
Deploy AI across the firm — and still answer "why did this number change?"
For CTOs, heads of technology and heads of risk, NAS is a single calculation standard: every desk and every AI tool calls the same engine and gets the same number, with the full trail behind each result.
Use NAS to:
- Establish one calculation standard across every desk and book
- Govern AI access to pricing and risk with agent identity, audit trails and permission scoping
- Reproduce and defend any number on demand — to a risk committee, an auditor or a regulator
- Run overnight batch on shared defaults with configurable local overrides, from pre-trade through end-of-day
- Eliminate key-person risk and library upkeep: Numerix maintains the engine; you own only the thin application layer
- Retire calculation infrastructure with a cloud-hosted, AWS-native, SOC 2-certified service
Powered by 30 years of Numerix analytics
NAS runs on Numerix's market-standard analytics libraries — three decades of financial engineering as best-practice, configurable defaults. It delivers dealer-quality coverage across equities, credit, interest rates, FX, inflation, fixed income (including MBS/ABS), commodities, hybrids and crypto, all behind a single high-level API, with more on the roadmap.
Bring your own market data, or use built-in reference and market data for supported instruments (tri-party Bloomberg and Refinitiv arrangements supported). Changes flow through robust CI/CD with nightly numerical regression, a Preview environment one to five weeks ahead of Production, and a backward-compatibility guarantee.
Proven, provable, in production
Ranked #1 by Chartis
NAS is built on the analytics platform Chartis named Overall Winner of its Quantitative Analytics50 2026. Numerix took top honors across Impact and Strategy, Innovation and Solution categories — including XVA, CVA, MVA, OTC derivatives, equity derivatives, integrated risk analytics and AI-driven asset pricing.
Frequently asked questions
What is NAS?
NxCore Analytics Services (NAS) is Numerix's cloud-hosted calculation engine, delivered as an API. It gives AI applications a deterministic foundation: the same inputs produce the same pricing and risk numbers every time, across every asset class, and any result can be reconstructed on demand.
Why does determinism matter if my AI is already accurate?
Accuracy on the happy path is not the issue. A probabilistic model can return a plausible number that no one can reproduce, defend or reconcile across desks — and you often discover that when a regulator or risk committee asks. NAS separates the two roles: the AI does the language, NAS does the math.
How does NAS work with AI and agents, and what is available today?
NAS exposes a high-level, financial-language API that LLMs read natively, with LLMs.txt documentation and an OpenAPI schema. Agents such as Claude call NAS through a standards-compliant MCP server, or integrate through the NAS Python API and REST API.
Which asset classes and data does NAS cover?
Dealer-quality cross-asset analytics across equities, credit, interest rates, FX, inflation, fixed income (including MBS/ABS), commodities, hybrids and crypto — spanning valuations, Greeks, P&L, VaR and counterparty credit risk including XVA. Bring your own market data, or use built-in reference and market data for supported instruments; tri-party Bloomberg and Refinitiv arrangements are supported.
Why NAS instead of building in-house or using a trading platform?
NAS integrates with your existing systems rather than requiring rip-and-replace, and Numerix maintains the engine so you avoid key-person risk and library upkeep. It is SOC 2-certified with a 99.9%+ uptime SLA, has run in production for six years. Unlike a black-box calculation, a diagnostics endpoint reproduces any result on demand.
How do I get access?
Schedule a walkthrough below — bring your hardest pricing question or your own term sheet and see it priced live, with the trail behind the number. A trial environment is available with no infrastructure setup; your Numerix representative can confirm the right tier for production use.