# NaxVerse

> Nyx is an AI-native decisioning engine for banks, fintechs, payment processors, lenders and merchants: decision and trust infrastructure for transactions initiated by a human or by an AI agent. It returns explained and audited decisions in real time, and extends through versioned domain packs. Fraud management and agent discovery are the two packs ready to pilot today, on the institution side and the merchant side respectively; agent trust is in design with design partners; credit risk, underwriting, disputes, AML and KYC compose on the same governed engine. It deploys three ways: hosted and managed by NaxVerse, in the customer's own cloud, or in the customer's data center. Built by NaxVerse, headquartered in Noida, India, working with customers in any jurisdiction. Pre-launch: a small design-partner cohort is forming.

Key facts, stated carefully:

- The problem, as three forces compressing the decision window at once. One, the attack is automated: deepfake KYC, synthetic identity and GenAI social engineering are generated faster than supervised models retrain (42.5% of fraud attempts AI-driven, Signicat 2024; $486B global fraud losses in 2023, Nasdaq Verafin 2024). Two, the liability is moving onto the party making the decision, on staggered timelines (EU PSD3/PSR liability shift from 2028; UK APP-fraud reimbursement in force since 7 Oct 2024). Three, the customer may not be human: agents find, evaluate and buy on a person's behalf, and the device and behavioral signals every stack leans on vanish with them (agentic payment rails live since 2025, Visa TAP and Mastercard Agent Pay; $1.5T agentic commerce by 2030, Juniper 2025; 25% of breaches from AI-agent abuse by 2028, Gartner 2024). Fraud is where most decision stacks break first, but none of the three stops at fraud.
- Why now: regulators worldwide are making AI fraud and automated decisions the institution's liability (EU Verification of Payee under the Instant Payments Regulation and PSD3/PSR liability shift from 2028; UK mandatory APP-fraud reimbursement up to GBP 85,000 in force since 7 Oct 2024; the US with no federal mandate yet but liability pressure building; the MAS Shared Responsibility Framework in Singapore; RBI Master Directions in force since July 2024 in India; Australia's Scams Prevention Act 2025; SAMA Counter-Fraud Framework in force since 13 Apr 2026 and CBUAE Notice 3057 fines live since 31 Mar 2026 in the Gulf). Nyx is regulator-neutral by construction and deploys in any jurisdiction; no market is a boundary.
- Who it is for: banks, digital banks, fintechs, payment processors, acquirers, lenders and merchants. Size is not a gate, because the hosted deployment removes the need for an in-house platform team. On the agentic-commerce pack the buyers are merchants, marketplaces and platforms, in the order direct-to-consumer brands, then smaller merchants, then marketplaces, acquirers and PSPs; that sequence is about where the first proof gets made, not about who NaxVerse will work with.
- Use cases: one engine, specialized by pack. Fraud management (pack built, pilot ready); agent trust and agentic commerce (agent discovery pilot ready, agent trust in design, design partners shaping it); credit risk and underwriting (in design, open to design partners); disputes and chargebacks, and AML, KYC and transaction monitoring (primitives in place, pack scoped with the first partner). Adjacent decisions such as collections, merchant onboarding risk, limit management and payment screening map onto the same primitives and are scoped on request.
- Agentic commerce has two directions, and Nyx addresses both from one canonical data model and one set of policy. Discovery, being found and understood: a customer shopping through an agent on a surface such as ChatGPT, Gemini or Copilot only sees businesses whose products, availability, capabilities (delivery, returns, financing, eligibility) and policy are readable by that agent, answered live from the business's systems rather than from a nightly export. A business that is unreadable is absent from the answer and never learns what it lost. This is the agent-facing layer, ready to pilot with the first design partners; the ingestion, canonicalization and policy machinery underneath already exists, and the new work is the publish surface. Trust, deciding what you accept: agents already transact on the card rails and arrive without the device, session and behavioral signals fraud models were trained on, carrying a mandate instead. Nyx verifies the agent credential the rails present, resolves the principal who delegated authority, evaluates delegated scope, per-transaction and cumulative caps, expiry and revocation in the authorization path, and returns an explained, audited verdict. This is in design. The two halves belong together because the eligibility rule that tells an agent who may buy is the rule that decides whether the purchase stands; split across two systems they disagree in public.
- Agentic boundaries, stated on the website: Nyx does not issue agent identity (the networks and protocols do, and Nyx decides on what they present); Nyx does not replace a commerce platform or a PSP, since the catalog of record stays where it is and money moves on existing rails; Nyx cannot promise placement or ranking inside any agent surface, because being readable by an agent is not the same as being chosen by one; agent-to-agent negotiation is stated as direction, not sold as product; nothing agentic is benchmarked, the agent-facing surface is ready to pilot, and agent and mandate fields are in design rather than shipped.
- What it does: one decisioning engine from data ingestion to decision. Synchronous scoring in the authorization path returns the action, score, severity band, policy and rule evidence, model version and audit reference in a single API response.
- Where it runs: three deployment models on the same codebase and the same API. Nyx-hosted (managed by NaxVerse in the region the customer chooses, single-tenant by default, multi-tenant where one institution serves its own merchants on it); the customer's own cloud account and VPC (AWS, Azure or GCP); or the customer's data center. Built on open standards (PostgreSQL, Redis, Kafka, Spark, MLflow, KServe, Keycloak, OpenTelemetry). Models train on the customer's data only, and there is no bulk data extraction.
- Performance: sub-100ms scoring, 30 to 50 percent fewer false positives, 45-second investigations and 3 to 5 times analyst throughput are DESIGN TARGETS to validate in the first pilot, not yet independently benchmarked. NaxVerse publishes targets, not testimonials it does not have.
- Rollout: designed to go live in weeks (a designed rollout model, to be proven with first design partners). Day 1 means day one of deployment, once the environment is provisioned.
- Team: built by a team with two decades across the payments lifecycle at JPMorgan, TSYS and Pine Labs.
- How Nyx is built: with NOX, NaxVerse's own development harness. One tracker ticket in, one reviewed pull request out, with scope planned up front, review and security gates that cannot be skipped, and a recorded trail for every change. NOX is in private beta in the NaxVerse tools library at https://get.naxverse.ai/. It is engineering governance, not a second product line; Nyx is the product NaxVerse sells.
- Company: NaxVerse Technologies Pvt Ltd, Noida, Uttar Pradesh, India.
- Contact: hello@naxverse.ai (pilots, walkthroughs, press, due-diligence pack).

## Pages

- [Home](https://naxverse.ai/): the executive brief: the three forces compressing the decision window, the six decision packs and their status, the engine, the three deployment models, the team and the pilot CTA
- [Agentic Commerce](https://naxverse.ai/agentic-commerce): the two halves (be discoverable by a customer's agent, then decide on that agent when it buys), who NaxVerse is starting with and in what order, the four layers with their honest build status, what a design-partner engagement involves, and an explicit list of what NaxVerse is not claiming
- [Decisioning for institutions](https://naxverse.ai/decisioning): the hub for all six decision domains, what each pack decides, who runs it, and how far along it is
- [Fraud management](https://naxverse.ai/fraud-management): the institution-side pack that is ready to pilot, what it decides, the two decision paths, the design targets and what a first pilot looks like
- [Credit risk](https://naxverse.ai/decisioning#credit): exposure, limit and pricing decisions on the same engine; pack in design, open to design partners
- [Underwriting](https://naxverse.ai/decisioning#underwriting): policy-driven approvals with the reasoning attached; pack in design, open to design partners
- [Platform](https://naxverse.ai/platform): architecture for technical evaluators, shared primitives, the decision API, deployment models, operational posture and governance
- [Privacy](https://naxverse.ai/privacy): what data the website and pre-launch conversations collect, and how it is handled
- [Terms](https://naxverse.ai/terms): terms of use for the website itself, separate from any pilot or deployment agreement

## Detail

- [Full detail](https://naxverse.ai/llms-full.txt): the complete machine-readable description