xyt, the independent trading data intelligence platform formerly known as big xyt, has introduced a set of AI-powered capabilities designed to let clients query its datasets in natural language, integrate its data into their own AI environments, and generate executable analytical outputs from a prompt. The announcement positions the firm in an increasingly crowded field of independent trading data providers racing to layer generative and agentic interfaces on top of their historical and real-time market data.
The new functionality, available to clients now, spans three areas. Users can explore xyt data through natural-language prompts, bypassing the query-writing and validation steps that traditionally sit between a trading question and a defensible answer. Clients building their own AI tools or internal research workflows can connect xyt data directly into those environments, although the company has not disclosed the underlying integration mechanism. And users can describe an analytical task in plain language and receive a complete output – data, visualisations and logic – that can be refined and operationalised in production.“We are enabling our clients to interact with data in a completely new way: faster and more intuitively, all without compromising on control or transparency,” says Robin Mess, CEO of xyt. “Our goal is simple: help clients get to better decisions, quicker. With these new enhancements, they can move from question, to insight, to action.”
Independence as positioning
The release lands against a backdrop of rapid consolidation among market data, trading analytics and consolidated tape providers. xyt, which rebranded from big xyt in September 2025 and secured a €10 million investment led by Finch Capital in late 2024, has explicitly staked out ground as an independent alternative to incumbents as the industry reshapes around acquisitions. Its Trading Data Intelligence Platform processes more than 12 billion messages daily across 120-plus global venues, covering pre-trade, intra-trade and post-trade use cases for hedge funds, banks, brokers, exchanges, asset managers, ETF issuers and market makers.
By adding AI-accessible interfaces on top of that data estate, xyt is targeting a demand-side signal that has become unmistakable over the past year: institutional clients want to plug trusted market data into their own LLM workflows, without committing to a vendor-defined dashboard or report format.
A pattern across the independents
xyt is not the first to make this move, and the competitive context matters. In March 2026, BMLL and Tradefeedr announced a partnership to build what they described as an AI-ready analytics layer for equities and futures, combining BMLL’s Level 3 historical order book data with an agentic interface Tradefeedr has built on top of its analytics APIs. Tradefeedr’s system – LLM-agnostic, supporting Claude, OpenAI and Google models – lets clients query trading data in natural language, with the orchestration layer of agent workflows, governance controls, vector databases and MCP integrations providing the differentiation.The parallels with xyt’s announcement are striking. Both firms are independent trading data specialists; both are pitching natural-language access as a displacement mechanism for traditional dashboards and reports; and both are framing the underlying data quality as the competitive moat, with the AI layer as the delivery mechanism.
What matters for clients is how these capabilities land in practice: whether the natural-language layer is tightly coupled to a pre-defined set of analytical workflows, or whether it exposes the underlying data flexibly enough for buy-side and sell-side teams to build their own agent-driven applications on top. That distinction – constrained assistant versus open data substrate – will shape which independent providers institutional firms choose to standardise on.
Institutional guardrails
xyt does, however, foreground the institutional requirements that any such capability has to clear. Every prompt and result, the company says, is transparent, traceable and reproducible, grounded in verified, standardised datasets so that AI-driven insights can be audited and embedded into regulated workflows. This is the non-negotiable layer for capital markets AI: without deterministic query translation, prompt logging and dataset versioning, natural-language access to trading data is a research toy rather than a production-grade tool.
It is also the layer where the competition will ultimately be decided. Most front-office AI initiatives that have reached production have done so by solving governance and explainability first and capability second. The firms that can credibly offer both high-fidelity trading data and an AI interface that survives a compliance review will have a material advantage as institutional adoption accelerates.
The interface battleground
Several questions will determine how significant this announcement proves to be. Is the external-environment integration a dedicated API, a Model Context Protocol implementation, or something else? How does the agentic workflow layer compare with Tradefeedr’s, and with the in-house tooling that sophisticated buy-side clients are already building on top of raw trading data? And which client segments – market structure research, TCA, product development, algo performance review – are driving the earliest adoption?
xyt’s strengths in liquidity analytics, execution quality measurement and consolidated tape services give it a distinctive base to build from. Whether this launch marks the start of a serious push into agent-accessible analytics, or remains a natural-language veneer over the existing platform, will become clearer as client deployments emerge. For now, the direction of travel across the independent trading data segment is unambiguous: the interface layer is becoming the new battleground, and xyt has made its opening move.
Subscribe to our newsletter
