Blog · AI
— AI··10 min read

How to brief a freelancer when AI is doing part of the work

Joona Heinonen· Choco Media · Rovaniemi

When AI is doing part of the work, the freelancer brief needs to say so — clearly, specifically, and without apology. Choco Media works with freelancers on a regular basis, and the single biggest source of rework we’ve seen is not skill gaps or bad briefs. It’s ambiguity about the AI content collaboration boundary: what the model produced, what the human is expected to transform, and what “done” actually looks like.

This post is for teams and agencies that already use AI in content production and now need freelancers to work within that system without guessing their role. If you’re briefing a copywriter, editor, translator, or content strategist on a job where a language model has done some of the heavy lifting, the standard brief format isn’t enough. You need a few extra fields — and this post gives you the template.

By the end you’ll have a practical brief structure you can adapt immediately, a clear framework for defining the AI/human split, and a set of language guidelines that prevent the most common friction points we’ve seen in mixed-workflow projects.

Why the standard freelancer brief breaks down in AI-assisted workflows

The classic content brief covers the topic, the audience, the word count, the tone, and a handful of keywords. That was enough when a freelancer was starting from a blank page. It’s not enough when they’re starting from a 1,400-word AI draft that’s 70% usable and 30% wrong in ways that aren’t obvious.

The problem is invisible assumptions. The client assumes the freelancer will “just polish it.” The freelancer assumes they’re doing a light edit. Neither assumption is stated. So the freelancer sends back a lightly edited version, the client receives something that still has hallucinated statistics and AI sentence structure, and everyone is confused about how it happened.

The fix is a brief that explicitly names the AI’s role, defines the human’s task, and sets quality expectations that account for both.

The eight fields every AI-assisted brief needs

These fields go on top of — not instead of — your standard brief. Think of them as a small “AI context block” you append to whatever format you already use.

1. AI involvement level

State clearly what the model produced. Use plain language: “This is a full AI draft (approximately 1,800 words). Your job is a substantial rewrite, not an edit.” Or: “The AI produced an outline and the intro paragraph. You’re writing the rest from scratch using the structure provided.”

Avoid vague labels like “AI-assisted” or “partially AI-written.” They tell the freelancer nothing actionable.

2. What to keep

Identify the parts of the AI output that are correct and should be preserved: the structure, specific examples, a particular explanation, the keyword placement. This saves the freelancer from throwing out good work and saves you from losing elements you specifically prompted for.

3. What to fix

Call out the known problems: “The statistics in section 3 are unverified — please replace with cited sources or remove.” “The tone becomes overly formal in the middle sections — bring it back to the voice reference.” “The final CTA is generic — rewrite to match the service page linked below.”

4. Voice and tone reference

Link to a published piece that represents the target voice. Not a style guide document — a real, live URL. Freelancers calibrate to examples faster than they calibrate to described rules. At Choco Media we link to posts like our Notion content engine piece as a voice reference when briefing editors on AI drafts.

5. Factual verification scope

Be explicit about what requires source verification. “All statistics, tool names, pricing, and feature claims must be verified and linked.” If you don’t say this, some freelancers will assume the AI’s confident phrasing means the facts were checked.

6. Output format expectations

Specify the deliverable format: clean HTML, Google Doc with tracked changes, plain text with semantic headings. If the post will go into a CMS with a specific structure (like our own renderer), say so and link to an example output.

7. What the AI was prompted to do

Include the original prompt or a summary of it. This gives the freelancer context for why the AI draft looks the way it does — what constraints were given, what angle was requested, what it was told to avoid. It also helps them spot where the model drifted from the brief.

8. Word count and scope of human work

State the expected output length and how much human transformation is required. “Deliver 1,800–2,000 words. Expect to rewrite approximately half of the AI draft and add three verified examples.” Vague scope leads to either over-delivery (unnecessary hours) or under-delivery (a light polish on content that needed surgery).

Setting the human/AI boundary clearly

The most useful thing you can do in the brief is name the boundary explicitly. We use a simple three-tier model:

The brief’s job is not to judge whether AI involvement is good or bad. It’s to give the freelancer an accurate picture of the work so they can price it, scope it, and do it well. Ambiguity here is a cost, not a time-saver.

In our experience, most mixed-workflow briefs fall into “draft transformation.” That’s the mode where the brief details matter most, because it’s the mode where the freelancer’s judgment calls have the biggest impact on output quality.

Language to use (and avoid) in the brief

How you describe the AI’s role affects how the freelancer approaches the work.

Use precise action verbs

Instead of “clean up the AI draft,” say “rewrite the argument in sections 2 and 4, verify all statistics, and adjust tone throughout to match the voice reference.” Instead of “make it sound more human,” say “break up sentences over 30 words, remove hedging phrases like ‘it’s worth noting that’ and ‘it’s important to consider,’ and replace abstract claims with specific examples.”

Avoid implying the AI draft is the standard

Phrases like “just improve what the AI wrote” or “the AI did the hard part” set the wrong expectations. They suggest the AI output is the baseline to build from, not a starting material to critically assess. In practice, accepting an AI draft as the baseline is how hedged, generic content gets shipped.

Be direct about quality expectations

If the post needs to be indistinguishable from fully human-written content by a skilled writer, say so. If a functional, well-structured piece is sufficient, say that too. The freelancer can deliver either, but only if you’re clear about which one you’re paying for. If you need AI-assisted content creation at scale, setting these expectations upfront is what makes the pipeline work.

How to handle revisions in AI-assisted workflows

Revision cycles in AI-assisted projects tend to go wrong at the same point: the client gives feedback that’s really feedback on the AI draft, not the freelancer’s work.

Before giving revision notes, check which elements were AI-generated versus freelancer-written. If your feedback is “this section sounds flat,” find out whether that section was in the original AI draft unchanged. If it was, the note belongs in the original brief as a constraint, not in the revision request.

This approach makes revisions faster and keeps the working relationship clear. Freelancers are more receptive to feedback when they understand what they’re being asked to improve and why.

Pricing and scope: what to tell the freelancer

The scope of work in AI-assisted projects is genuinely different from a full write-from-scratch commission, and the pricing should reflect that — in both directions.

A “draft transformation” pass on a 2,000-word AI draft typically takes 60–90% of the time of a full original write, depending on the quality of the AI output and the level of fact-checking required. A “quality pass” on a well-reviewed AI draft might take 20–30%. Neither is free, and neither should be priced as if it’s the same as writing from scratch.

When briefing, give the freelancer enough information to estimate accurately. Include: the approximate word count of the AI draft, the transformation tier (structural / draft transformation / quality pass), the fact-checking scope, and the number of revision rounds included.

If you’re running a content operation that uses AI at scale, building a standard rate card for each transformation tier is worth the hour it takes. It removes negotiation friction on every brief and gives freelancers the predictability they need to prioritize your work. This is the kind of operational detail that sits within a broader AI automation workflow — getting the human-AI handoff right is as important as getting the prompts right.

A brief template you can use today

Here’s the structure we use when briefing freelancers on AI-assisted content. Adapt the headings to your own format.

  1. Project overview — topic, target audience, platform, publication date
  2. AI involvement level — structural / draft transformation / quality pass
  3. AI draft file — link to draft + original prompt summary
  4. What to keep — specific sections, examples, or structures to preserve
  5. What to fix — known errors, tone problems, gaps to fill
  6. Voice reference — URL of a published piece in the target voice
  7. Factual verification scope — what must be checked and sourced
  8. Output format — format, length, heading structure, delivery method
  9. Revision rounds — how many included, turnaround expectations
  10. Rate and scope confirmation — agreed rate for this tier, what triggers a scope change

Running this brief format reduces rework consistently. The reason is simple: it eliminates the gap between what the client imagines and what the freelancer attempts. That gap is where most mixed-workflow projects lose time.

What this looks like in practice

In client work we’ve found that the first AI-assisted brief is always the hardest. Teams default to their existing brief format, add a note that “AI was used,” and hope the freelancer figures out the rest. The second brief is better because at least one revision cycle has clarified what was missing.

The brief template above is designed to skip that first cycle. It’s not complex — ten fields, fifteen minutes to fill out. But it forces you to answer the questions the freelancer would otherwise have to ask (or guess).

The payoff compounds. Once a freelancer has completed one well-briefed AI-assisted project with you, they calibrate quickly. They know your AI output quality, your voice standards, your verification expectations. The brief becomes a checkpoint, not a discovery document.

If you’re scaling content production — whether for your own brand or for clients — the human/AI handoff is one of the few places where a small investment in process design returns disproportionate quality improvement. Getting the brief right is that investment.

If you want to talk through how we structure mixed-workflow content operations, reach out to us directly — we’re happy to walk through what’s worked and what hasn’t.

— Work with Choco Media

Want posts like this working for your business?

10–40 SEO + AI-optimised blog posts a month, researched, senior-edited and published straight to your site. Built to rank on Google and get cited by ChatGPT, Claude and Gemini.

See plans — from €199/mo →
No start-up fee · Price locked for 12 months · Cancel any time after
← All storiesNext story →
— Free tips, monthly

Get the playbook, for free.

One short letter a month — the prompts we use, the campaigns that worked, the AI tools worth the time. No sales pitch, just field notes.

— Want us to do it for you?

Hire the agency.

AI-accelerated content, paid media, brand and web — delivered by one small team that talks to itself. Currently taking on a handful of clients each quarter.

Book a call