Much of what legal labels “complexity” is really administrative burden. AI makes it possible to isolate the judgment, redesign the process and stop using expensive expertise to find problems a system could have surfaced first.
A former colleague once told me there was no such thing as complex legal work.
I thought he was being ridiculous.
Our teams were sized around complexity. Work was routed according to complexity. Seniority, pricing and delivery models all depended on whether a matter had been classified as low, medium or high complexity.
I spent a few weeks trying to prove him wrong. Eventually, I went back and admitted that he had a point.
That does not mean legal work never requires sophisticated judgment. It means that we often use one word to describe two very different things.
There is intellectual complexity: genuine ambiguity, competing commercial priorities, novel law, and decisions that require experience and accountability.
Then there is operational complexity: long documents, fragmented information, dense drafting, repetitive review, inconsistent processes and large amounts of manual effort.
The distinction can be uncomfortable. Complexity carries status: lawyers build expertise and professional identity around handling difficult matters.
AI is becoming a great leveler here, making it possible to reduce—and in some cases remove—the operational burden we have traditionally treated as complexity. The need for judgment remains. The need for hours of manual effort increasingly deserves a challenge. The document is not always the difficult part.
Consider a standard commercial agreement.
Someone has to read it, compare it with the organization's preferred position, identify deviations, prepare a summary, suggest revisions and surface the issues that require a decision.
Historically, that work was allocated as one package to a lawyer. Because the overall matter carried risk, every step inherited the label "complex."
But the steps are not equally complex.
Reading every clause is not the same as deciding whether a liability position is commercially acceptable. Finding a deviation is not the same as deciding whether to concede it. Producing a summary is not the same as balancing legal and business risk.
AI can support much of the reading, comparison, classification and issue-spotting when it is given the right standards and context, and its performance has been tested for the task.
The judgment has not disappeared. It has become easier to locate.
AI can also assist with intellectual complexity: helping lawyers compare interpretations, explore the implications of different positions and surface assumptions. The ambiguity may remain, but the options become clearer.
The opportunity is to design the workflow so each step receives the right level of attention, with legal expertise focused where it adds the most value. If one clause in a 70-page agreement requires a senior lawyer, it does not follow that the same lawyer needs to spend hours locating that clause, reading standard provisions and preparing the materials required to make the decision.
The default should no longer be "send it to a lawyer"
Most legal workflows still begin with assignment.
A request arrives. Someone decides who is available. The document is sent to a lawyer, and only then does the real work begin.
AI makes it possible to reverse that sequence.
The workflow can start by asking:
What type of agreement is this?
Do we have all the information and business context needed to assess it?
How closely does it fit the standard position?
Is there a commercial, regulatory or value-based trigger?
The system can then create a first-pass review pack before senior legal resource is involved: a summary, risk assessment, relevant obligations, playbook deviations, proposed redlines and the issues requiring judgment.
By the time the lawyer sees the work, they should have a clearer view of the likely issues and the supporting evidence.
Their job includes validating that view and deciding what to do about it. They should not have to reconstruct the entire matter simply to get started.
From doing the work to designing it
Lawyering becomes less about who does every task and more about who designs the work.
Simply adding AI to an existing workflow is not enough. We also need to choose the right tools for each task and redesign how the work is delivered, avoiding unnecessary AI use and wasted effort. This is the principle of disaggregation: break the work into its component tasks and match each task to the right person, location and technology.
Legal expertise also becomes part of the delivery infrastructure: the playbooks, review standards, escalation rules and reusable methods that help an entire team work consistently. That is a huge responsibility. It does require letting go of the idea that personal involvement in every step is the thing that makes the work valuable.
A better question about complexity
The legal sector will not stop using the word "complex." Nor should it.
But the definition needs to become more precise.
AI does not make every legal problem simple. It makes it possible to stop treating every part of the process as equally difficult.
That is the real opportunity.
Not replacing expertise, but redesigning legal work so expertise is used where it matters most.
Talk to the Factor team about redesigning your contracting workflows so AI reduces routine effort and your lawyers can focus where judgment matters most.
