What is the most common mistake when using AI to write?

Giving it too little information and then trying to edit the vague output into shape. It is faster to spend one minute writing a specific brief than ten minutes fixing a response that guessed what you needed.

How do I get AI to write in my own voice?

Paste two or three examples of your previous writing and ask the model to match that style. Then add the specific constraints that distinguish your voice: sentence length, vocabulary level, phrases you use or avoid. Iterate on a few short examples before applying the style to a full piece.

Can AI write something long, like a full report?

Yes, but section by section works better than asking for the whole thing at once. Write a brief for each section, review it, and build the full document from approved sections. This also makes it easier to maintain consistency across sections.

How do I make sure the AI does not include wrong facts?

Supply the facts in your brief and tell the AI not to add information beyond what you gave it. For factual writing, always verify specific claims, names, dates, and statistics before using them. AI models can generate plausible-sounding but incorrect details.

Do I need a paid AI subscription to do this?

The free tiers of most AI apps are enough for most writing tasks. Paid plans typically offer longer context windows, which helps when you are working with a long document. Start with what you have and upgrade if you run into context length limits.

How to Use AI to Write Something You Can Actually Use

A beginner's guide to getting usable written output from AI: how to frame a request, what to include, and how to review and refine the result.

October 15, 2026Updated October 21, 20265 min read

Using AI to write something useful means giving the model the information it needs before asking it to write. A request that specifies the audience, the purpose, the length, and the constraints produces a usable first draft. A vague request produces a generic one that needs so much editing it would have been faster to write from scratch.

The steps below work for any kind of writing: emails, reports, blog posts, project updates, or study notes. Tools like fenbs keep your writing context and your best prompt versions in one place so you build on what worked rather than starting over each time.

Step 1: define the audience and purpose

Before you open a chat, answer two questions: who will read this, and what should they do or think after reading it? "Write a project update" is a task. "Write a project update for a client who does not know our internal process, telling them the key milestone we hit and the one decision they need to make" is a brief.

The AI cannot make this decision for you. If you leave the audience undefined, it will guess, and it usually guesses a general, educated adult who wants a balanced overview — which is not always what you need.

Step 2: give it the facts it needs

A writing AI does not know your project, your client, or your organization. Supply the facts. Include the key point to communicate, any background the reader needs, and the specific details that belong in this piece.

Brief template

Write a [type of writing] for [audience]. Purpose: [one sentence]. Key facts to include: [list]. Facts to avoid or not mention: [list]. Length: approximately [word count or "one paragraph"]. Tone: [formal / friendly / direct / etc.].

You do not need to fill every field for every piece of writing. For a casual internal update, audience and key facts may be enough. For a formal document, all of them matter.

Step 3: show it an example if you have one

If you have a piece of writing that is close to what you want — a previous email in the right tone, a report in the right format — paste a short excerpt and say "write in a similar style to this." A model calibrates to a concrete example more reliably than to a description of style.

If you do not have an example, describe the style in terms of what you want to avoid. "No bullet points, no bold headers, write in flowing paragraphs" is more useful than "professional but warm." Negative constraints are often easier to specify and more reliably followed than positive ones.

Step 4: review the draft against your request

Read the draft against the request you wrote, not against a general sense of whether it reads well. Did it include the key fact? Did it speak to the right audience? Is it the right length? These checks tell you specifically what to ask for in the next version.

A common mistake is to start editing the draft yourself instead of sending a second prompt. Editing takes longer, and the result is a mix of AI text and your own that can feel inconsistent. A focused follow-up prompt — "remove the third paragraph and cut the conclusion to one sentence" — gets you a clean draft in seconds.

Resist the urge to accept a draft that is mostly right. A second prompt — "shorten the second paragraph and remove the list" — takes less time than editing the output yourself, and the third version is usually much closer to what you need.

Step 5: save what worked

After a session where the AI produced something genuinely useful, note what made it work. Was it a specific way you framed the audience? A constraint that cut the padding? An example you pasted in? Saving that version of the prompt means you start from a working template next time rather than rebuilding from scratch.

How fenbs helps

Keep your writing context and best prompts in one place

fenbs Skills and prompts stores your finalized templates as named entries. If you write project updates every month, save the version of your request that worked. A connected assistant reads it at the start of the next session and applies it without you re-explaining the audience and constraints. The writing context — who the stakeholders are, what the project is — lives in the project record.

Create a free account

Free for one person · No credit card · Works with Claude, ChatGPT, Cursor and more

How to do this with fenbs

Create a free account and set up a project for the work you write about regularly.
In the project, add context notes with the facts that belong in every piece of writing for this project: audience, key background, constraints.
In Skills and prompts, save your best brief template for each type of writing you do regularly.
Connect an AI app from Integrations. At the start of a writing session, ask it to read the project context and the relevant skill.
After finishing a piece, note any decisions about style or format in Decisions and rules so they apply next time automatically.

fenbs is not a writing app or editor. It keeps your brief and decisions so each writing session starts with the right context. Your finished documents live wherever you usually keep them — a shared drive, a document app, an email thread — and fenbs keeps the record of what was decided and what to do next.

If you are new to AI writing, the how to write a task for an AI agent article covers a related skill: describing work precisely enough that an agent can carry it out. That framing applies to any work you hand to an AI, not just writing tasks.