A New Frontier: Early Data Strategy in the Age of Generative AI
Author: LDI Team
A midsize financial services and investment firm received a formal Securities and Exchange Commission (SEC) inquiry. The legal team scrambled to collect and review the data but couldn't meet the SEC's deadline for producing the requested information.
Ultimately, the firm was cleared of any wrongdoing. But it still had to pay a penalty for missing the deadline. The regulator's position was straightforward: An organization of this size and sophistication, operating in a regulated industry, should have known where its data was.
What really led to the delay?
There was no data map, no preservation protocol, and no documented inventory of which systems held what data. The legal team had no idea who the custodians were or any inkling of where their data resided.
No generative AI tool, no matter how sophisticated, could have yielded the information the SEC required because the firm's data estate was essentially ungoverned.
The delay came down to one thing: the legal team didn't know its own data until a moment of urgency.
From Early Data Assessment to Early Data Strategy
In October 2025, LDI Architects published a white paper defining Early Data Assessment (EDA) as the systematic process of identifying, organizing, and analyzing data at the earliest stage, often before a formal matter has begun, using insights from historical data to help drive legal strategy.
Since then, the use of generative AI has expanded even further. Its capabilities have simplified data analytics, making it easier to generate reports and dashboards that once required teams of analysts and engineers.
Taking account of these fast and seismic changes in the technology landscape, the LDI Architects for Disputes and Investigations published a new guide that explores how legal teams can move beyond EDA, into the next frontier: Early Data Strategy (EDS).
"From Assessment to Intelligence: Operationalizing Early Data Strategy (EDS) in the Age of Generative AI," was co-authored by LDI Architects Kevin Clark, Melina Efstathiou, Ross Gotler, Matt Hamilton, Tristan Jenkinson, Glenn Melcher, Daniel Miller, and Linda Sheehan.
Guide
From Assessment to Intelligence: Operationalizing Early Data Strategy (EDS) in the Age of Generative AI
Get the GuideEDS is about generating intelligence early, turning the data landscape that EDA maps into forward-looking strategy. Where EDA asks what data exists and what it might cost or the risks it might carry, EDS asks what the data reveals and what to do about it. Its focus areas move from inventory to analysis: early data analytics and visualization, issue and fact-pattern identification, witness and document linkage, damages and exposure modeling, and scenario analysis and strategy testing. Its outcomes move accordingly, from a data map to actionable insight, a stronger case strategy and a better decision, made earlier.
If EDA is a road map, EDS is a compass.
First, Know Thy Data
One practical example of EDS in action is a general counsel querying his organization's legal portfolio in natural language and then receiving a reliable answer. Not a dashboard built by an analyst, and not a report commissioned for a single purpose, but a direct, real-time response: the highest-frequency claim types over the last 24 months, the cost drivers behind each, outside counsel performance across the portfolio, the matters that share a risk profile with an emerging dispute.
That capability requires generative AI. But as the example of the financial services firm that was penalized by the SEC illustrated earlier, you can't truly unlock the benefits of generative AI unless the underlying data has been carefully governed.
EDS requires everything generative AI needs to function reliably in a legal context: governed data, accurate taxonomy, consistent intake, reliable access controls, and a clear ownership structure.
Read "From Assessment to Intelligence: Operationalizing Early Data Strategy (EDS) in the Age of Generative AI" if your organization is adopting generative AI faster than it is governing the data the generative AI depends on, because it spells out concretely what legal teams must do before any of the promised intelligence becomes reliable or defensible.
The guide helps you build a foundation for EDS, walking you through steps such as curating your data so redundant, obsolete, or trivial (ROT) data doesn't corrupt outputs; mapping what data you hold, where it lives, and which AI systems can access it on what terms; scrutinizing vendor contracts for training-data and data-residency provisions legacy agreements never contemplated; and connecting information, AI system, and decision governance atop a single current data map.
It also features a section titled "War Stories," contributed by the LDI Architects, which details case studies of organizations that paid penalties, over-collected documents, or lost years of trial time for one reason: they didn't know their data before they needed it.
Check out "From Assessment to Intelligence: Operationalizing Early Data Strategy (EDS) in the Age of Generative AI" to learn more.