You've read about AI for a year. Have you built anything yet?
You’ve followed newsletters and attended webinars, but if asked to create something with AI right now, would you know how to begin?
August 17, 2026
You have followed the newsletters. You have sat through the webinars. You could probably explain what a large language model does at a dinner party. And if someone asked you to open a blank document right now and build something with AI, today, would you know where to start?
For a lot of professionals in our community, the honest answer is no. Not because they are behind. Because they have spent a year collecting information and almost no time practising with it.
That gap has a name worth naming: the knowing-doing gap, the space between being informed and being capable. It opens up for ordinary reasons. Reading feels like progress and building feels risky. Waiting for the right tool feels safer than picking an imperfect one. None of that is a character flaw. It is just a habit worth breaking, and breaking it is smaller than it sounds.
Five steps to build something this week
1. Pick one task you already have to do. Not a hypothetical project or a "someday" idea. A real email, LinkedIn post, report summary, or client brief that is already on your list this week. Using a live task means you will actually finish it, and you will have a genuine reason to judge whether the output was any good.
2. Choose one tool and stop comparing. ChatGPT, Claude, Copilot, Gemini. Any one of them will do for a first attempt. The tool matters far less than the practice. If you are already paying for one through work, start there and save the comparison shopping for later.
3. Write a prompt with four things in it: context, audience, goal, and one example. A vague prompt like "write me a LinkedIn post about AI" will get you a vague result. Instead, try something closer to: "Write a LinkedIn post for a marketing audience about why AI has changed how we brief content, not whether we still need writers. Keep it under 150 words. Here is a post of mine from last month so you can match the tone: [paste example]." Specific instructions are what separate a usable draft from a generic one.
4. Read the output like an editor, not a customer. Do not accept the first draft, and do not throw it out either. Mark what is genuinely useful, cut what does not sound like you, and rewrite the opening line yourself if the tool's version is flat. The editing is not a chore to get through. It is the actual skill you are building.
5. Do it again within the week, on a different task. One attempt tells you whether the tool works. A second attempt, on something different, tells you where you personally tend to under-specify, and what kind of output you can trust without a heavy rewrite. That pattern recognition is what separates someone who has "tried AI once" from someone who can use it on demand.
Three mistakes that stall this process
Treating the first draft as the finished product. AI output is a starting point, not a submission. Publishing it unedited is how people end up with content that sounds generic, and it is usually the reason someone decides "AI doesn't work for me" after one bad experience.
Writing prompts the way you'd talk to a search engine. Three or four words rarely gets you anywhere useful. Treat the prompt like a short brief you would hand a junior colleague: what is this for, who is it for, and what does good look like.
Practicing on a task with no real stakes. A fake exercise is easy to abandon halfway through. A real deadline, even a small one, is what forces you to actually finish, edit, and use the result.
Why this works better than more reading
Competence comes from repetition on real tasks, not from a wider list of tools you understand in theory. Each of the five steps above takes under an hour. Do all five once and you will have produced something real, with your own judgement built into it, rather than another bookmark.
The confidence that follows is not a mindset shift. It is the direct result of having done it and seen that the output, once edited, was actually usable.
Take the next step
If you want structured practice rather than another list of tools to bookmark, our AI Content Mastery course is built around exactly this process: real briefs, real editing, real feedback. It moves you from reading about AI content to producing it.
Written by the She Loves Data editorial team
You've read about AI for a year. Have you built anything yet?
You’ve followed newsletters and attended webinars, but if asked to create something with AI right now, would you know how to begin?
August 17, 2026
You have followed the newsletters. You have sat through the webinars. You could probably explain what a large language model does at a dinner party. And if someone asked you to open a blank document right now and build something with AI, today, would you know where to start?
For a lot of professionals in our community, the honest answer is no. Not because they are behind. Because they have spent a year collecting information and almost no time practising with it.
That gap has a name worth naming: the knowing-doing gap, the space between being informed and being capable. It opens up for ordinary reasons. Reading feels like progress and building feels risky. Waiting for the right tool feels safer than picking an imperfect one. None of that is a character flaw. It is just a habit worth breaking, and breaking it is smaller than it sounds.
Five steps to build something this week
1. Pick one task you already have to do. Not a hypothetical project or a "someday" idea. A real email, LinkedIn post, report summary, or client brief that is already on your list this week. Using a live task means you will actually finish it, and you will have a genuine reason to judge whether the output was any good.
2. Choose one tool and stop comparing. ChatGPT, Claude, Copilot, Gemini. Any one of them will do for a first attempt. The tool matters far less than the practice. If you are already paying for one through work, start there and save the comparison shopping for later.
3. Write a prompt with four things in it: context, audience, goal, and one example. A vague prompt like "write me a LinkedIn post about AI" will get you a vague result. Instead, try something closer to: "Write a LinkedIn post for a marketing audience about why AI has changed how we brief content, not whether we still need writers. Keep it under 150 words. Here is a post of mine from last month so you can match the tone: [paste example]." Specific instructions are what separate a usable draft from a generic one.
4. Read the output like an editor, not a customer. Do not accept the first draft, and do not throw it out either. Mark what is genuinely useful, cut what does not sound like you, and rewrite the opening line yourself if the tool's version is flat. The editing is not a chore to get through. It is the actual skill you are building.
5. Do it again within the week, on a different task. One attempt tells you whether the tool works. A second attempt, on something different, tells you where you personally tend to under-specify, and what kind of output you can trust without a heavy rewrite. That pattern recognition is what separates someone who has "tried AI once" from someone who can use it on demand.
Three mistakes that stall this process
Treating the first draft as the finished product. AI output is a starting point, not a submission. Publishing it unedited is how people end up with content that sounds generic, and it is usually the reason someone decides "AI doesn't work for me" after one bad experience.
Writing prompts the way you'd talk to a search engine. Three or four words rarely gets you anywhere useful. Treat the prompt like a short brief you would hand a junior colleague: what is this for, who is it for, and what does good look like.
Practicing on a task with no real stakes. A fake exercise is easy to abandon halfway through. A real deadline, even a small one, is what forces you to actually finish, edit, and use the result.
Why this works better than more reading
Competence comes from repetition on real tasks, not from a wider list of tools you understand in theory. Each of the five steps above takes under an hour. Do all five once and you will have produced something real, with your own judgement built into it, rather than another bookmark.
The confidence that follows is not a mindset shift. It is the direct result of having done it and seen that the output, once edited, was actually usable.
Take the next step
If you want structured practice rather than another list of tools to bookmark, our AI Content Mastery course is built around exactly this process: real briefs, real editing, real feedback. It moves you from reading about AI content to producing it.
Written by the She Loves Data editorial team
You have followed the newsletters. You have sat through the webinars. You could probably explain what a large language model does at a dinner party. And if someone asked you to open a blank document right now and build something with AI, today, would you know where to start?
For a lot of professionals in our community, the honest answer is no. Not because they are behind. Because they have spent a year collecting information and almost no time practising with it.
That gap has a name worth naming: the knowing-doing gap, the space between being informed and being capable. It opens up for ordinary reasons. Reading feels like progress and building feels risky. Waiting for the right tool feels safer than picking an imperfect one. None of that is a character flaw. It is just a habit worth breaking, and breaking it is smaller than it sounds.
Five steps to build something this week
1. Pick one task you already have to do. Not a hypothetical project or a "someday" idea. A real email, LinkedIn post, report summary, or client brief that is already on your list this week. Using a live task means you will actually finish it, and you will have a genuine reason to judge whether the output was any good.
2. Choose one tool and stop comparing. ChatGPT, Claude, Copilot, Gemini. Any one of them will do for a first attempt. The tool matters far less than the practice. If you are already paying for one through work, start there and save the comparison shopping for later.
3. Write a prompt with four things in it: context, audience, goal, and one example. A vague prompt like "write me a LinkedIn post about AI" will get you a vague result. Instead, try something closer to: "Write a LinkedIn post for a marketing audience about why AI has changed how we brief content, not whether we still need writers. Keep it under 150 words. Here is a post of mine from last month so you can match the tone: [paste example]." Specific instructions are what separate a usable draft from a generic one.
4. Read the output like an editor, not a customer. Do not accept the first draft, and do not throw it out either. Mark what is genuinely useful, cut what does not sound like you, and rewrite the opening line yourself if the tool's version is flat. The editing is not a chore to get through. It is the actual skill you are building.
5. Do it again within the week, on a different task. One attempt tells you whether the tool works. A second attempt, on something different, tells you where you personally tend to under-specify, and what kind of output you can trust without a heavy rewrite. That pattern recognition is what separates someone who has "tried AI once" from someone who can use it on demand.
Three mistakes that stall this process
Treating the first draft as the finished product. AI output is a starting point, not a submission. Publishing it unedited is how people end up with content that sounds generic, and it is usually the reason someone decides "AI doesn't work for me" after one bad experience.
Writing prompts the way you'd talk to a search engine. Three or four words rarely gets you anywhere useful. Treat the prompt like a short brief you would hand a junior colleague: what is this for, who is it for, and what does good look like.
Practicing on a task with no real stakes. A fake exercise is easy to abandon halfway through. A real deadline, even a small one, is what forces you to actually finish, edit, and use the result.
Why this works better than more reading
Competence comes from repetition on real tasks, not from a wider list of tools you understand in theory. Each of the five steps above takes under an hour. Do all five once and you will have produced something real, with your own judgement built into it, rather than another bookmark.
The confidence that follows is not a mindset shift. It is the direct result of having done it and seen that the output, once edited, was actually usable.
Take the next step
If you want structured practice rather than another list of tools to bookmark, our AI Content Mastery course is built around exactly this process: real briefs, real editing, real feedback. It moves you from reading about AI content to producing it.
Written by the She Loves Data editorial team
