Are Paralegals in Demand? What the 2035 Projections Show

Short answer: demand is flat, not falling, and the composition of the work is changing faster than the number of jobs. The US Bureau of Labor Statistics projects paralegals and legal assistants to show little or no change between 2025 and 2035, a decline of about 1,100 positions from a 2025 base of 404,900, with a 2025 median wage of $62,890. Against 3.5 percent growth for total employment, flat is below average and a long way from collapse. Meanwhile lawyers are projected to grow 5 percent, adding 40,600 positions, which tells you the legal sector is expanding while the support layer inside it is not.

Flat Is the Interesting Number

A projection of essentially zero change over a decade is unusual and more informative than either growth or decline would be.

It means two forces are cancelling out. Legal work is increasing, pushed up by regulation, transactions, disputes and compliance. At the same time the share of that work needing a person is falling. Document review, discovery processing, contract abstraction, citation checking and standard drafting have all been heavily automated.

For an individual, that average hides a wide spread. If your week is document handling, the role is shrinking. If it is case management, client contact and litigation support, it is stable or growing. The occupation number is those two trends cancelling, which is why it looks like nothing is happening.

What Has Been Automated, and What Has Not

The rows at the bottom are where the role is consolidating. Anything involving a person, a deadline and a consequence stays human, and anything involving processing a stack of documents does not.

The verification problem is new work

There is one genuinely new category of paralegal work, and it exists because of AI rather than despite it.

Generated legal text is fluent, well-structured and cites authorities with confidence. Some of those authorities do not exist. Others exist but do not say what they are cited for. Courts have taken a dim view of filings with fabricated citations, and the professional consequences land on the firm.

Checking every cited case is real, still good law, and actually supports the proposition attached to it is precise, tedious, high-stakes work. It needs both legal research skill and an understanding of how these systems fail. That is a paralegal task. It did not exist three years ago, and most firms discover they need it after an incident rather than before one.

A matter that shows where the value sits

Follow one litigation matter to see why the projection is flat rather than falling.

Discovery produces 400,000 documents. Ten years ago a team of paralegals and contract reviewers would have worked through that for weeks. Today a review platform narrows it to perhaps 8,000 potentially responsive documents in a day, and that step is genuinely gone.

The paralegal’s work starts after that. Someone has to set the review protocol and defend it if the methodology is challenged. Someone has to work the documents the system flagged as ambiguous, which are disproportionately the interesting ones. Someone has to make privilege calls on borderline material, and getting that wrong carries consequences no software absorbs.

Then the matter continues, and none of it stops for the technology. Deadlines have to be tracked across jurisdictions with different local rules. A filing gets rejected on a formatting technicality at 4pm and has to be corrected before the clerk closes. A witness needs preparing and their availability changes twice. During a deposition the partner needs one specific document, and needs it in under a minute.

None of that runs itself.

None of that is document processing, and all of it is paralegal work. The volume task disappeared and the matter still needs running, which is the whole explanation for a projection that moves neither up nor down.

What to Know Before You Draw Conclusions

Supervised practice rules create a floor. Paralegals work under attorney supervision and cannot give legal advice. That structure means the responsibility for output stays with a licensed person, which is precisely the arrangement automation does not change.

Firm economics are shifting from hours to outcomes. Once document review stops being billable at volume, pressure moves to everything else. If your value was measured in pages processed, you are exposed. If it is measured in matters kept moving, you are not.

Practice area matters enormously. Litigation support, immigration, family law, personal injury and real estate closings all involve high client contact and procedural complexity, which is what keeps a person in the loop. Corporate document work sits at the opposite end and is the most exposed part of the profession.

Exposure measures are not employment forecasts. BLS published AI exposure categories with the 2025-35 projections and states plainly that exposure “does not imply job loss, productivity gains, automation probability, or wage effects.” Legal support work scores high on task overlap and the projection is flat rather than negative.

In-house teams are hiring differently from firms. Corporate legal departments handling contract volume increasingly want paralegals who can run and audit an automated contract process, which is a different job description from the traditional one.

Where Paralegal Demand Is Actually Growing

• Litigation support and e-discovery management. Someone has to run the process, defend the methodology, and handle the material the system flagged ambiguously.
• Compliance and regulatory support. Growing obligations across sectors create steady work with real procedural complexity.
• Immigration. Document-heavy and deadline-driven, with high client contact and consequences for error that make automation a support tool rather than a substitute.
• Trial preparation. Exhibits, witnesses, logistics and the ability to find the one document a partner needs in eight minutes.
• AI output verification. The new category described above. Most firms have not staffed it and discover the need after an uncomfortable incident rather than before one.
• Legal technology operations. Configuring, testing and governing the tools the firm bought. In most firms nobody formally owns this, which makes it the easiest responsibility to claim and one of the hardest to replace once you hold it.

A Decision Framework for Paralegals

1. Mostly document review or contract abstraction. The most exposed position. Move toward case management, client contact or litigation support within the next year, and treat the flat projection as a warning rather than reassurance.
2. Litigation paralegal. Comparatively secure, because trial work is logistics and judgement under pressure. Your growth move is owning the e-discovery process rather than working inside it.
3. In-house or corporate paralegal. Your differentiator is becoming the person who runs and audits the automated contract workflow, which is a promotion in most departments rather than a sideways step.
4. Considering entering the profession. Enter through a practice area with client contact and procedural complexity rather than a document-processing role, and expect verification skills to be part of the job from day one.

The test that applies to all four: how much of your week could be described precisely enough for someone to reproduce it from written instructions? That share is your exposure. The rest, which usually involves a client, a deadline or a judgement call, is the durable part.

Common mistakes right now

• Assuming a flat projection means nothing is changing, when it conceals two large opposing shifts.
• Accepting generated research or citations without verifying every authority independently.
• Competing on document processing speed, which is now measured against software.
• Neglecting the technology the firm has bought, when being the person who understands it is the clearest available differentiator.
• Treating client communication as administrative overhead rather than as the part of the job that no tool replaces.

Building the Verification Skill Firms Now Need

The most valuable thing a paralegal can learn right now is not a new practice area. It is a precise understanding of how generated text fails, because that understanding is what turns an automated draft from a liability into a genuine time saving.

Concretely, that means four things. Why fabricated citations look exactly like real ones. Why a summary can be fluent and wrong in a way that survives a quick read. What must be checked independently rather than spot-checked. And how to fit a verification step into a workflow under deadline pressure. Learning that in a structured sequence beats acquiring it from an incident, and it makes you the person a supervising attorney trusts with the output. If you want a structured route in, explore Coursiv AI lessons and check current plan details on the official site.

FAQ

Your Next Step

Take last month’s timesheet and split it into work that processed documents and work that moved a matter forward. If the first column dominates, that is your exposure in a single number, and the flat projection means you have time to act rather than a reason to relax. Then volunteer for the thing nobody in your firm currently owns, which in most firms is checking whether the output of the tools they bought is actually correct.