Author: Lincoln Wang | Founder of MindsLeap | Global Partner at Founders Space | Founder of Founders AI Club
"People's trust in AI begins with trust in data"
This didn't come from an AI startup. It came from the London Stock Exchange Group (LSEG).
LSEG sits at the core of global financial markets. The information it processes every day directly affects large-scale capital flows and critical business decisions. In this context, AI is not a tool for writing emails — it is infrastructure embedded in product DNA, directly shaping client judgment.
LSEG has built a deep partnership with OpenAI, feeding trusted data and services into ChatGPT via MCP (Model Context Protocol). But this is not simply "pouring data into a model." When LSEG evaluates every AI product, it looks beyond data accuracy. It examines groundedness — whether answers can be traced back to evidence — the model's reasoning logic for financial information, data redundancy, and data fidelity.
In finance, a single wrong number can trigger a chain reaction. LSEG's approach reveals a sequence that is often overlooked: the first step for enterprises adopting AI is not choosing a model — it is confirming whether your data is worthy of being used by AI.
Two-week release cycles, built on a rebuilt data foundation
LSEG revealed a telling detail: their product release cycle used to be three to six months, depending on the product and complexity. Now, all AI product releases are compressed to two weeks.
Biweekly iteration is nothing new in software. But inside an institution that manages global financial information infrastructure, this pace means the entire data supply chain has been redesigned.
Information flows faster. Decisions move faster. Iteration follows suit. This is not because models got smarter — it is because the underlying data models have been realigned, rebaselined, and made consumable by AI at scale.
Many enterprises complain about inconsistent AI output quality. The problem is usually not on the model side — it is on the data side. What you feed AI has not been systematically cleaned, aligned, and annotated. LSEG's clients did the heavy lifting first: adjusting data models, resetting baselines, building scalable consumption methods. The model is only the last mile.
Analyst time, returned to research
LSEG highlighted a notable shift: the role of analysts is expanding dramatically.
Previously, financial analysts spent most of their time on data preparation, basic information verification, and report generation. Now that AI handles these tasks, analysts can pursue deeper research and uncover more "orthogonal" insights — the kind of work they never had time for before.
This is not an "AI replaces humans" narrative. It is an "AI reshapes how humans spend their time" narrative.
When basic information processing is automated, human expertise no longer shows up in "who organizes fastest." It shows up in "who asks better questions" and "who spots connections others miss." For every knowledge-intensive industry, the signal is clear: redesigning roles is not about headcount reduction — it is about unlocking professional capacity trapped in low-value work.
27,000 people: fast and safe at the same time
LSEG has 27,000 employees worldwide. The challenge they face mirrors that of large Chinese enterprises: how do you get all of them to embrace AI without losing control?
Their answer: make scaling best practices easier, while embedding the necessary standards and skills to ensure quality floors are never breached.
Here is a key point that is easy to miss. Many assume "scaling AI" means distributing tools to everyone and hoping value emerges naturally. LSEG does the opposite — standards first, skills embedded, quality guardrails in place, then scale.
Speed and safety are not opposites. They are sequential. Speed without guardrails is an accident waiting to happen. Speed with guardrails is a genuine competitive advantage.
Trusted data as new infrastructure
LSEG's move to feed trusted data into ChatGPT via MCP has implications that may not yet be fully appreciated.
When enterprise users access LSEG's financial data directly inside ChatGPT, they are not buying "AI capability." They are buying a combined experience of trusted information plus AI interaction. The models themselves are increasingly commoditized. But verified, fidelity-checked, AI-consumption-aligned data is becoming the new scarce resource.
The takeaway for Chinese entrepreneurs is direct: the quality of your data assets determines how far you can go in the AI era. Anyone can call a model. Your proprietary data, the digitalization of your industry know-how, the rigor of your information architecture — these are the moats that cannot be easily replicated.
Back to the starting point: the order of trust cannot be reversed
The entire LSEG–OpenAI narrative points to a simple but frequently ignored judgment:
The enterprise AI investment sequence should be: data trust → product guardrails → scaled distribution → accelerated iteration. Too many companies reverse this order — rushing to get everyone using AI first, only to discover the underlying data cannot withstand scrutiny and AI output cannot be directly adopted by business teams.
Releasing AI products every two weeks sounds like an efficiency story. But what supports that efficiency is all the unglamorous work that came before: data model alignment, fidelity verification, reasoning logic review. None of it is sexy, but all of it determines whether AI can move from demo to production.
For Chinese enterprises planning their AI strategy, this may be the most important question to pause on: Is your data worthy of being used by AI?
If the answer is not yet clear, the first step is not finding a better model. It is going back to the data itself.
About MindsLeap
MindsLeap is an AI-native organization transformation accelerator.
In deep partnership with Silicon Valley innovation incubator Founders Space, MindsLeap connects cutting-edge global AI insights, the Silicon Valley tech entrepreneurship ecosystem, and real-world transformation scenarios for Chinese entrepreneurs.
MindsLeap is building a transformation ecosystem for entrepreneurs, startup founders, AI engineers, industry experts, and investors — helping enterprises move AI from awareness, strategy, and tools into organizational capability, business processes, product innovation, and growth systems.
This article was translated and adapted from the Chinese original with AI assistance.