Legal Engineering: Two Years In

Norm Ai’s mission is to make legal and compliance advice abundant by building Legal AGI. To achieve this mission, we needed a way to automate and scale human legal expertise so that we could embed it into our AI systems. In 2024, Norm Ai created a new discipline, Legal Engineering, to make this possible.

Legal Engineers are lawyers who build, calibrate, and supervise AI agents that perform legal analysis at scale. Two years in, the discipline of Legal Engineering has matured faster than anyone anticipated. In this article, we’ll walk through that evolution.

Central to this growth is how we train Legal Engineers. Attorneys who join Norm undergo a multi-week internal training program that culminates in a certification. Until they are certified, Legal Engineers cannot begin building client-deliverable AI products. The curriculum covers everything from the fundamentals of how large language models work to hands-on development with Norm's proprietary tools. The curriculum itself changes on a monthly basis as AI capabilities evolve. What we teach a new Legal Engineer today is materially different from what we taught six months ago.

Building the Logic (2024)

When the first Legal Engineers joined Norm Ai, they took on ambitious work given where frontier models were at that time: translating dense regulatory texts into structured, interpretable logic that AI agents could execute. Using Norm Ai’s proprietary Legal Engineering Automation Platform (LEAP), attorneys began embedding statutes, regulations, and legal workflows directly into AI systems, building granular logical frameworks with hundreds of underlying components.

This was a fundamentally different kind of legal work from law firm practice. Rather than producing static memos or interpretive guidance for a single client, Legal Engineers were constructing decision architectures: reusable, scalable representations of regulatory logic. Each one had to be precise enough for a machine to execute and nuanced enough to reflect the complexity of the underlying law.

This approach quickly became the underlying engine used in live compliance reviews by asset management firms managing trillions of dollars.

Rigorous Evaluation and Client Specificity (Late 2024 – Mid 2025)

The next phase was about making these agents adaptable to the specific needs of individual clients, each with their own regulatory interpretations, risk appetites, and internal guidelines, while maintaining the rigorous consistency that clients require.

Norm’s Ai and software engineers continued to improve the internal tools available to Legal Engineers. This new generation of tooling, leveraging the continued improvements of frontier models, enabled Legal Engineers to configure each regulation to each client’s preferences with extreme granularity. For instance, a single subsection, such as SEC Rule 206(4)-1(a)(5) requirement for specific investment advice to be “fair and balanced,” could operate with a “base” and “client specific” layer of analysis. This pattern repeated across every subsection of every regulation in the Norm Ai platform.

Empirical validation became central to the workflow. Legal Engineers began systematically measuring and proving agent performance, including designing evaluations to compare results across various frontier models and client contexts.

Legal Engineers Start Writing Code (Late 2025 into Early 2026)

The pace of AI tooling development accelerated dramatically in the second half of 2025. Norm’s model, in which proprietary tools built by our software engineers sit on top of external AI infrastructure, allowed us to incorporate these advances rapidly and equip Legal Engineers with the capabilities to build legally viable workflows at a pace that would not have been possible even months earlier.

Attorneys who had never written a line of code began building functional software tools: web applications, AI agents with email integrations, and internal workflow automations. They did this using cutting-edge AI coding tools, including Norm’s proprietary coding tools built specifically for legal work. Legal Engineers were demoing working prototypes to senior-level clients within days of conceiving them. This unlocked both dramatically faster production of core regulatory analysis and entirely new categories of output: tools, interfaces, and capabilities that had not existed before.

Legal Engineers now work in terminals, use command-line interfaces, and build with the same development environments used by software engineers. Norm's software engineers build the tools that enable Legal Engineers to do this, creating an integrated model where legal expertise and engineering capability compound on each other.

Building and Supervising Agents (April 2026)

Legal Engineers now think in terms of agent workflows, not individual tasks. They build chains of agents that handle increasingly complex legal analysis, set objectives for those agents, generate code, evaluate results, and drive refinements. They spend their time applying high-leverage judgment to the agents working on their behalf, standing behind the work product those agents produce.

This applies in two directions. Internally, Legal Engineers use agents to automate and accelerate their own workflows, building credible, defensible outputs faster than was previously possible. Externally, Norm's Ai agents supervise real-world work product and outcomes for clients and for the practicing attorneys within Norm Law, LLP.

Context and supervision are what distinguish genuine legal infrastructure for AI agents from surface-level automation. As AI agents take on increasingly autonomous roles across industries, the need for expert legal oversight of their outputs becomes foundational. Legal Engineering is the discipline that provides it, and it is core to what we believe is the emergence of Legal AGI: AI systems capable of performing sophisticated, contextual legal reasoning across domains, built and supervised by lawyers who understand both the law and the technology.

Attorney-Driven Legal Engineering at Norm Law, LLP

Over the past two years, Norm Ai developed a proven process for training attorneys to become Legal Engineers. Lawyers from top law schools and firms underwent our intensive retraining, learning to harness large language models, understand the boundaries of frontier AI, and build domain-specific AI agents.

That same training methodology is now applied to practicing attorneys at Norm Law, LLP, our affiliated law firm. Norm Law attorneys bring the rigor and obligations of active legal practice to the work of building and supervising AI agents, while drawing on the infrastructure and tooling that Norm Ai’s Legal Engineers have built. This partnership between Norm Law and Norm Ai effectively creates the ultimate R&D lab for Legal AGI: practicing attorneys and Legal Engineers collaborate on a daily basis to design and deploy AI agents that perform substantive legal work. Legal engineering is not only a product discipline. It is now underpinning legal practice within a full-service, AI-native law firm.

Where Legal Engineering Is Heading

The frontier models will get better. The scope of what a single Legal Engineer can own will continue to expand as coding tools become more sophisticated. Over the coming months, the “engineering” in Legal Engineering will become even more salient. The core skills that define Legal Engineering will endure: judgment in identifying which problems to solve, intuition for how cutting-edge AI capabilities can be marshaled toward a solution, and expertise in building and supervising agents that execute credible legal reasoning and create economic value. Norm Ai is building the legal infrastructure for AI agents, and Legal Engineers are the people who make that infrastructure reliable, adaptable, and worthy of trust.

Disclaimers

Legal Engineers are non-practicing lawyers employed by Norm Ai who design and build the AI systems that support Norm Law’s service delivery. Neither Norm Ai nor its personnel (including Legal Engineers) provides legal advice or services.