Most disappointing AI output comes down to one thing: a thin brief. Ask a tool to “write a blog post about commercial litigation” and it will, but the result reads like every other AI-generated page on the same topic, because it has nothing specific to work from.
The fix is not a better tool. It is a better brief. Treated properly, briefing AI is closer to briefing a junior colleague than typing a search query, and the same information a colleague would need to do the job well is exactly what AI needs too.
Why the brief matters more than the prompt
There is a tendency to treat AI prompts as something to get right through trial and error: tweak the wording, add “professional tone,” try again. This can improve results at the margins, but it rarely fixes generic output, because wording was never the problem.
The problem is usually missing information. AI fills gaps with the most statistically likely content, which for legal marketing means broad, safe, forgettable phrasing. The way to avoid that is to leave fewer gaps. A good brief does this by covering four things: context, audience, outcome and constraints.
Context: what AI needs to know before it can help
Context is the background a colleague would already have from working at the firm or chambers, and that AI has none of unless it is told. This includes what the piece is actually about, why it is being written now, and any relevant facts, cases, or developments driving it.
For a chambers or law firm, useful context might include the specific practice area or type of work involved, any recent case, ruling or regulatory change prompting the piece, how this content fits with existing pages or previous posts, and any facts, figures or examples that should be reflected.
Vague context produces vague drafts. “Write about employment law changes” gives AI almost nothing to differentiate on. “Write about the practical impact of the upcoming changes to fire-and-rehire rules for HR teams in the retail sector” gives it something to actually say.
Audience: who the content is for
The same topic reads very differently depending on who it is for. A briefing note for in-house counsel needs different depth, tone and vocabulary than a piece aimed at members of the public trying to understand their rights.
Worth specifying: who is reading this (in-house counsel, instructing solicitors, the public, journalists), how much they already know about the subject, and what they are likely trying to work out by reading it.
Chambers content, in particular, is often written for a professional audience of solicitors and clerks rather than the public. AI defaults to a general audience unless told otherwise, so without this detail, it will tend to over-explain basic concepts and under-deliver on the specifics that a professional reader actually wants.
Outcome: what success looks like
AI cannot judge whether a draft has succeeded unless it knows what success looks like. This is not the same as the topic. It is the job the content needs to do.
Useful outcome detail includes what the reader should think, feel or do differently having read it, where the content will be published or sent, and roughly how long it should be.
“Write a page about our commercial disputes team” is a topic. “Write a page that helps an in-house counsel decide whether this set has the right experience for a mid-value shareholder dispute” is an outcome. The second version gives AI a test to write towards.
Constraints: the boundaries that keep drafts usable
Constraints are what stop a technically fine draft from being unusable in practice. For legal content, this usually covers claims that cannot be made without evidence, house style requirements such as British English, sentence case, or no calls-to-action, any regulatory or professional conduct considerations, and length or format limits.
Legal marketing carries more risk than most sectors if constraints are skipped. AI left unchecked will readily produce lines like “our team of experts provides comprehensive legal solutions,” phrasing that overstates expertise and cannot be substantiated. Stating constraints upfront, rather than editing them out afterwards, saves a full pass of corrections later.
Using examples effectively
Examples do more work than most other parts of a brief, because they show rather than describe. Pointing AI to a previous piece of content in the house style, a page that gets the tone right, or a competitor’s page that illustrates what to avoid, gives it something concrete to match against.
The most useful examples are specific rather than general. “Match the tone of this LinkedIn post” works better than “make it sound professional,” because “professional” means something different to everyone, while a real example removes the ambiguity.
It is also worth showing AI what not to do. A short note such as “avoid the kind of generic opening used in this draft” is often more effective than several lines of instruction, because it points directly at the pattern to avoid rather than describing it abstractly.
Avoiding generic output
Generic AI output has a recognisable shape: broad claims that are not backed by specifics, empty superlatives like “leading” or “trusted,” and structure that could apply to almost any firm or chambers in almost any practice area.
The antidote is specificity, at every stage of the brief. Naming the actual practice area rather than “legal services.” Referencing a real case, statistic or example rather than a general assertion. Describing the actual reader rather than a generic professional. Stating what makes this firm or chambers different rather than leaving AI to invent a differentiator, which it will do with confidence and no basis.
A brief built this way leaves AI very little room to default to filler, because every gap that filler would normally occupy has already been filled with something real.
The result of a proper brief
None of this guarantees a publishable first draft. AI-generated content still needs review, fact-checking and editing before it goes anywhere near a website or client. What a proper brief does is change the starting point, from a generic draft that needs rewriting to a usable one that needs refining.
Given the choice between spending five minutes writing a better brief or twenty minutes rewriting a poor draft, the brief is the better use of time, every time.
Frequently asked questions
What is the single biggest reason AI produces generic legal marketing copy?
Missing context. AI fills in whatever is not specified with the most common, statistically likely phrasing, which in a legal setting tends to be broad claims and safe, forgettable language. Providing specific detail about the topic, audience and desired outcome removes the need for AI to guess, and guessing is where generic output comes from.
How long should a brief for AI actually be?
Length is not the goal, relevance is. A short brief that includes the right context, audience and outcome will usually outperform a long one that pads out vague instructions. The aim is to include the specific detail a colleague would need to do the job well, not to write at length for its own sake.
How much legal detail should be included in a brief for marketing content?
Enough to keep the content accurate, without expecting AI to make legal judgements. Facts, case references, and regulatory context that are directly relevant should be included. Legal interpretation, positioning, and any claims about the firm or chambers’ standing should still come from a person with the appropriate knowledge, and should be checked before publication regardless of how the content was drafted.
Can the same brief be reused for multiple pieces of content?
Parts of it can. Audience and constraints, such as house style and regulatory considerations, often stay consistent across a firm or chambers’ content. Context and outcome need to be specific to each piece, since these change depending on the topic and the purpose of that particular page or post
Is briefing AI different from briefing a copywriter?
Less than it might seem. A capable copywriter would ask similar questions before starting: who is this for, what should it achieve, what should it avoid saying, and are there examples to work from. The main difference is that a copywriter will ask when information is missing, while AI will not. It proceeds regardless, filling gaps with generic assumptions rather than flagging them, which is exactly why the brief needs to be complete from the outset.
