Legal engineers are not necessarily developers or traditional lawyers. Their work lives in the space between – where legal reasoning becomes workflow, and real-world user feedback informs how AI workflows, prompts, and product capabilities are iterated and improved.

From law firms to legal engineering

Mattia’s background in international tax and M&A shapes his view of legal AI because these areas involve a granular review of large volumes of documents to extract specific information, check consistency, and support tax-driven decisions. Based on that experience, he sees AI not as a replacement for legal judgment, but as a practical way to quickly extract relevant data from documents and automate repetitive, low-value tasks, allowing lawyers to focus on analysis, risk assessment, and decision-making. 

After several years in legal practice across Italy and the UAE, mostly focused on tax, he noticed a recurring pattern. “Much of the work was manual and repetitive, leaving too little room for the parts that truly require legal judgement,” he says. While tools existed, the technology at the time wasn’t mature enough to meaningfully address the problem at scale.

Eventually, Mattia joined a VC fund investing in early-stage AI startups. That’s where he came across Legora. When the legal engineer role opened up, he recognised it instantly. “It was the first job I’d seen that combined everything I’d been looking for.”

What legal engineers actually do

When Mattia describes his work now, he frames it simply. His goal is to make AI usable for lawyers without forcing them to become technologists. That means spending time with clients to understand how they actually work, and shaping the technology around those habits rather than asking lawyers to change overnight.

At Legora, legal engineers work across onboarding, tailored demos, product feedback, and workflow design. They act as a bridge between users and product teams, translating real-world legal needs into clear product requirements, share feedback loops with engineers, and help shape improvements that make the platform more accurate, usable, and aligned with how lawyers actually work. The role demands deep legal understanding, but also the ability to spot patterns. “You need to know what junior lawyers are doing every day,” Mattia says, “and what they shouldn’t still be doing.”

In tax, that often means structuring templates, setting up prompts for recurring reviews, or designing workflows that handle the repetitive groundwork. Human oversight remains central. The aim is not to remove lawyers from the process, but to focus their attention where it matters most.

Why tax is the ultimate AI use case

Tax law is full of structured logic, such as thresholds, classifications, and step-by-step qualifications. “It maps perfectly to workflows and decision trees,” Mattia explains. AI can support every part of the sequence - fact-gathering, rule selection, and documentation – without replacing professional judgment.

Generative AI, however, has clear limitations, especially when it comes to advanced mathematics and precise tax calculations, which are common in tax work. These limitations are addressed through a lawyers-in-the-loop approach: from the law firm side, where professional judgment remains essential, and through the support of Legora’s legal engineering team, which helps firms design prompts, workflows, and guardrails that reflect current LLM capabilities and constraints.

Building guardrails, not guesswork

Trust remains one of the biggest hurdles for AI in law, and tax raises the stakes even higher. “You can’t afford creativity in tax outputs,” Mattia says. That is why traceability is vital. At Legora, every output is grounded in source documents or verified data, with citations that allow users to follow each statement back to its origin. If something looks wrong, the path back to the source is immediate.

“It makes the review practical,” he explains. “You can double check what it found way faster, because Legora brings you straight to the right section of the doc.”

The cultural shift: from skepticism to strategic use

Adoption, however, is as much cultural as it is technical. Junior lawyers often see the value quickly because AI reduces repetitive work. For senior lawyers, the value can be less obvious but equally meaningful.

“For partners, AI is often more useful for analysis,” Mattia explains, such as extracting patterns across large datasets or turning complex information into clearer client insights. “Once they see that, it clicks.”

A new skillset for a new kind of lawyer

Looking ahead, Mattia sees a shift in emphasis rather than a complete change. Applying legal rules has always been essential. What matters now is the ability to adapt advice to a client’s specific context and structure problems in a way AI can support.

His advice to junior lawyers is simple: focus on how you think. Strong legal fundamentals still matter, but AI can help lawyers learn faster and apply that knowledge more effectively.

Not every lawyer will become a legal engineer. Deep legal expertise and sound judgment will remain essential. But as AI-powered workflows become more common, the differentiator will increasingly be a lawyer’s ability to use AI to learn faster, adapt advice to specific client contexts, and work more effectively at scale. In that landscape, legal engineers are likely to become an integral part of both law firms and in-house teams—translating legal expertise into advanced, AI-powered workflows and helping legal teams turn technology into practical impact.