Legal opinions provide the certainty markets need, but reliance on static, complex PDFs is rapidly becoming a barrier to progress. Marc Gratacos, Founder and Managing Director of TradeHeader, takes a look at how shifting to machine-readable "smart" opinions can finally bridge the gap between legal text and digital automation.
First published in TabbForum on December 03, 2025
On 19 November 2025, a collection of industry leaders gathered in London for the Smart Legal Opinions Symposium. The event wasn’t just another talking shop for the financial sector. The objective was specific and ambitious: to explore how we can move away from heavy, static documentation and towards “smart legal opinions” – real-time, machine-readable assessments that could fundamentally transform derivatives, digital assets, and financial markets.
For years, the industry has relied on established methods to handle legal certainty. However, the consensus from the room was that the old ways are struggling to keep up with the pace of modern finance. The path forward involves turning legal text into structured data, enabling a level of automation and trust that simply isn’t possible with today’s tools.
To understand the solution, we have to look at the problem. Legal opinions are the pillars of trust in the derivatives market. Specifically, netting opinions provide the necessary assurance that institutions need to transact safely across borders and with various counterparties. It is no exaggeration to say that without these opinions, the modern OTC derivatives market could not function.
However, the current state of these documents presents significant challenges. Over the decades, complexity has crept in. The English law netting opinion alone is now larger than all global legal opinions combined were in the 1990s.
As the volume of text has exploded, readability has plummeted. These documents have become increasingly difficult to interpret, validate, and maintain. Much of this is driven by the sheer complexity of financial regulation; in England alone, there are 20 different bankruptcy regimes, and each one must be analysed and addressed within these opinions.
A major hurdle discussed at the symposium was “semantic complexity”. Legal language is famous for its nuance and ambiguity – qualities that are useful in a courtroom but disastrous for automation. Jurisdiction-specific subtleties make it incredibly difficult to deploy intelligent tooling or standardisation.
Consider the scale of the maintenance task: ISDA currently publishes over 90 legal opinions every year. Reviewing and analysing this volume manually is inefficient, yet applying AI to the problem isn’t a magic bullet. While ISDA has introduced an AI Query tool to help navigate these documents, its impact is constrained by the unstructured nature of the source material.
We often hear about the potential of AI, but reliable systems require structured, high-quality data. If the input is messy or inconsistent, the output will be too. Identifying these gaps and low-quality inputs is becoming a critical requirement impacting when we can move forward.
The most promising solution discussed in London was the move towards greater structure. By introducing standardised formats and machine-readable schemas, we can enable AI tools to interpret opinions reliably. This would produce outputs that legal, risk, and operations teams can actually trust.
There are three concrete opportunities to make this happen:
One of the most effective ways to clear the fog is to change the language itself. Using “Basic Logical English” – a controlled, simplified form of the language – makes it readable and unambiguous, suitable for conversion into machine-executable representations. Standardising the language reduces the chance of interpretive disagreements and strengthens our ability to automate processes.
The Common Domain Model (CDM) already provides standardised lifecycle and data models for core industry documentation, such as ISDA CSAs and the ISLA Master Agreement. Extending the CDM to cover legal opinion structures is a natural progression.
By building structured, domain-specific models for legal opinions, the industry could codify legal logic consistently across different jurisdictions. This would enable intelligent search and validation while supporting automated netting and collateral workflows. Perhaps most excitingly, it would facilitate future smart contract implementations.
The ultimate goal is to evolve from static legal opinions – which are often 200+ page PDFs – to smart legal opinions. These would be semantically precise, machine-readable, and validated through standardised logic. Crucially, they would be able to operate in real-time.
The implications of this shift extend far beyond traditional derivatives. These capabilities would be transformative for digital assets, which are now formally recognised as property under UK law.
In a digital market, real-time legal reasoning is foundational. It is the key to enabling automated collateral management, real-time settlement, and programmable legal workflows. You cannot have a high-speed digital asset ecosystem relying on analogue, paper-based legal checks.
The symposium highlighted a shared belief among industry leaders: the tools, models, and standards required to build smart legal opinions already exist in some form. We do not need to invent new technology; we need to apply what we already have more intelligently.
What is needed now is the standardisation of language, the structuring of legal content, and the integration of these structures into industry standards like the CDM. By combining deep legal expertise with digital modelling discipline, we can move towards a future where legal certainty is automated, real-time, and interoperable across all asset classes.