AI Fluency in Legal Hiring and How to Show it

AI fluency is fast becoming a hiring priority for law firms. It means knowing how to use AI effectively in legal work, recognize its limitations, and apply your own judgment to its outputs. For students thinking about recruitment, that raises a practical question: What exactly do firms want students to know regarding AI, and how do you show you’ve got it? A few firms are leading the way with clarity around expectations. One recruitment post I came across recently opened with a single question: “Are you a lawyer who builds?”

The post didn't come from a legal tech start-up or an alternative legal service provider, but a newly launched AI-forward practice set up by a leading European law firm. To explain what it meant, the firm pointed candidates to Levels 4 and 5 of Harvey’s recently published AI Fluency Framework: Builders, who create reusable AI tools such as agents and playbooks, and Catalysts, who build for others and set the standards around them.

From AI Familiarity to Clear Expectations

A few years ago, it would have been unusual to see law firms recruit explicitly for Builders and Catalysts. In the case of the aforementioned recruitment post, the firm made it clear that it needs lawyers who create with AI, not just use it, and it used the framework's levels to outline their expectations. This example shows how the ideal candidate profile has changed and what law firms are looking for specifically — not just an introduction to AI but the ability to build agents and think critically about how and when to use AI, pointing to a clear framework to attain and measure those skills

If firms and candidates share a common language for AI fluency, the framework works just as well at the hiring stage as it does for a junior lawyer in development inside the firm.

A firm built around AI is likely to be a step ahead of the wider market, but its hiring expectations are an early sign of where recruiting is heading, and quickly. Among the most progressive firms, AI fluency is already a skill set in rising demand. Expectations are getting clearer — and higher. For students, that makes it worth understanding those expectations now, before they become the norm.

The Problem With "Comfortable With AI Tools"

Look beyond the most AI-forward employers, and the picture is less precise. Many legal job descriptions mention AI in some form: "comfortable with AI tools," "tech-savvy," "keen interest in legal innovation." These phrases signal a firm's intent, but they set a tone without setting a bar.

Comfortable doing what?

• Using AI to summarize a contract, identify key obligations, and check the summary against the source?
• Using AI to analyze deposition transcripts, identify inconsistencies, and verify findings against the record?
• Building an agent that teammates or a practice area can rely on, with clear instructions and review steps?
• Spotting where AI could reshape how a whole team works, then guiding changes to workflows and working practices?

These are very different levels of fluency. Today every role can benefit from a clear standard, and for most lawyers that standard should be Proficient: regular, confident use of AI on substantive legal work. Where firms need Builders or Catalysts, being specific matters even more.

The same clarity helps throughout a lawyer's career. A shared standard gives interviewers a consistent way to assess AI capability, gives development plans a clear destination, and shows both current and future lawyers what their goals should be, how to develop, and what their next step looks like.

What's missing is a common reference point: a shared understanding of what's expected, where the baseline sits and how to progress.

Five Levels of AI Fluency

Earlier this year, TITANS created The AI Fluency Framework, independent research commissioned by Harvey. It draws on interviews with legal technology and enablement leaders at seven law firms and four in-house teams across the UK, US, and continental Europe. At its centre is a five-level scale describing how an individual lawyer applies AI to legal work:

• Level 1 – Dormant: effectively a non-user who can't yet see where AI applies to their own work.
• Level 2 – Functional: uses AI occasionally for basic tasks like summarising, simple drafting and document querying.
• Level 3 – Proficient: uses AI regularly and confidently on substantive legal work, iterating and verifying for reliable output.
• Level 4 – Builder: builds reusable AI assets, such as agents and playbooks, for their own work and close colleagues.
• Level 5 – Catalyst: builds AI assets for a team, practice area or firm (or orchestrates the building) and sets the standards around them.

Each level comes with detection signals: what shows up in usage data, and how a lawyer talks and works. Designed to measure fluency across existing lawyers, they work just as well for describing what firms expect from new hires, and for helping students and lawyers understand the standard they need to reach.

For Students and Lawyers: Showing Your Fluency

For law students and junior lawyers, a stated level turns a vague expectation into a target.

If a firm says it expects Proficient associates, you know what to work towards: regular, confident use on substantive work, with verification built in. At this level, AI stops being something you remember to use and becomes part of how you work every day. If a role calls for a Builder, the bar is different. You'll need to show you can create something reusable, not just use AI well.

The detection signals also help you show your level. Rather than listing tools on a resume or cover letter, describe what you did with them:

• Show Proficient with a specific example of substantive work where AI played a real part, and the techniques you use to verify AI output.
• Show Builder with something you built for your own work or close colleagues, such as an agent or playbook. Explain the process it captures, the steps you made explicit, and what it improved.
• Show Catalyst with impact beyond your own circle: assets adopted across a team or practice area, standards or guidance you set, or colleagues you trained and mentored.

Not everyone has had the chance to build. Those permissions vary between firms, and students may not have access to enterprise tools. In that case, coursework, clinics, or personal projects still count, and so does being able to explain how you would approach building something: the process, the steps, and the checks.

The framework also works in reverse. Our research found that the biggest barriers to the Builder level are often environmental rather than individual: build permissions, time credited for experimenting, and governance foundations. Candidates can ask about these. Does the firm let lawyers build? Is time spent learning and experimenting recognised? Is there a path from using AI to creating with it? The answers will tell you a lot about whether a firm will help you grow.

How to get There

Knowing the target is only half of it. The framework's learning curriculum sets out the topics that carry a lawyer to each level, designed to hold across products rather than teach any one tool.

The Harvey Law School Program gives students opportunities to build AI fluency with tools designed for legal work. Whatever tools you use, the foundations come first: what AI is and where it falls short, hallucinations and bias, data security, policy, prompting, grounding, and verification. These topics are tool-agnostic, and a solid grounding in them is one of the clearest ways to show you’re ready to use AI on legal work. From there, the rest of the curriculum builds towards each level:

• To reach Proficient: effective review (using AI to review AI output), using AI as a thought partner, optimising outputs for the end reader, matching each task to the right feature, choosing the right model, chaining features across a task, and supervising work produced with AI.
• To reach Builder: a process mindset (inputs, steps, outputs), knowing when to build and what is worth automating, best practice for agents and reusable workflows, building within risk and governance rules, and testing and maintaining what you build.
• To reach Catalyst: a product mindset and ownership of shared assets, building at scale up to firm-wide, maintaining shared assets, setting standards and governance, writing best practice and guides, and coaching and mentoring others.

Beyond Practice: Legal Engineering

Builder and Catalyst skills aren't only valued in practice. They sit at the heart of legal engineering: roles at law firms, alternative legal service providers and legal technology companies that design, build, and maintain the way legal work is done with technology. Process thinking, reusable workflows, testing what you build and helping others adopt it are all core to that work. For students who find they enjoy building with AI more than a traditional legal career would allow, legal engineering is a route worth knowing about.

For Firms: Saying What You Expect

The first step is deciding where the bar sits for each role. Our research found a clear pattern across the organisations we spoke to: Proficient as the target for everyone, with Builders and Catalysts following as objectives and capacity require.

In practice, that might look something like this:

Expected levels may vary depending on the firm and the role, but the language for describing them doesn't have to. "C1 French" means the same thing to an employer and a candidate because a detailed standard sits behind it. Naming a level against a recognised framework, such as "we expect associates to be Proficient, as defined in the AI Fluency Framework," can create the same common ground, from the job ad to the interview and beyond.

The second step is assessing it. When a firm measures the fluency of its own lawyers, usage data does much of the work. Candidates don't come with usage data, so the signals need to become interview questions instead. The behavioural descriptions in the framework translate well:

• Proficient lawyers treat AI as a thought partner on the substance of their work, not just a drafting shortcut. A good interview question is a walk-through of a recent piece of substantive work where AI played a part. Strong answers show iteration, verification, and the right tool chosen for the task.
• Builders are process-minded. They see work as inputs, steps and outputs, and talk about processes rather than documents. A useful question is which task in a current role could become a repeatable workflow, and how. Strong answers start with the goal, the input sources and the steps, before reaching for a tool.
• Catalysts are product-minded. They build AI assets for others to rely on across a team, practice area or firm, and set the standards around them. Colleagues often consult them before building or sharing anything widely, and they mentor others along the way. Questions here might explore how to decide whether a colleague's agent is ready to share across the firm, and how to bring a sceptical team on board.

These questions don't depend on any one platform. A candidate from a firm using a different tool, or a student with no access to enterprise legal AI at all, can still show how they think. For students, these are also a good guide to the questions they may face.

The third step is carrying the same language beyond hiring. A level stated in a job ad becomes the baseline in a development plan and a milestone in progression conversations. Lawyers hear one consistent message from interview to promotion.

Keeping Expectations Realistic

None of this means every lawyer needs to reach the top of the scale. Expectations are rising, but they should reflect what each role requires. Not everyone needs to build, and not everyone wants to. For most lawyers, Proficient is a worthy destination in itself. Expecting every hire to be a Catalyst would be unrealistic and unnecessary.

Getting those expectations right means matching the level to the role, then supporting people to reach it. The firms doing this best are clear about the baseline they expect from everyone, deliberate about where they need Builders, and honest about the environment those Builders need.

Where to Start

For students and lawyers: Place yourself on the five levels honestly, decide which level the roles you want are likely to need, and gather one concrete example that shows it. Then pick the curriculum topics for the next level up.

For firms: Pick two or three roles you are hiring for now and decide which level each one needs. Name that level in the job ad and describe what it looks like in practice. Linking to the framework is optional, but it gives candidates the same reference point the firm is using. Then add one or two level-specific questions to the interview.

The full AI Fluency Framework report sets out the five levels in detail, including the detection signals and learning curriculum for each one.

Law students preparing for summer associate recruiting may also find Harvey's webinar on the law firm hiring process useful, covering how firms review applications and what interviewers look for.