Start with one useful fact, explain what it changes for a particular reader, and ask AI to draft within that evidence. Cut repeated points and claims you cannot support. Keep the details and phrasing that sound like you. Before publishing, check that the edit preserved the original meaning, numbers and qualifications.
What should a product post say first?
Lead with the task your reader recognizes or the result your product makes possible. “We shipped a major update” is news about the company. “Find overdue invoices by client” tells a freelancer what the update lets them do. The second sentence gives the reader a reason to care before asking for attention.
Choose one point per post. A feature release, a customer observation and a founder lesson each need different evidence. Putting all three into a short draft usually leaves too little space to explain any of them.
What should you give AI before it writes?
- A specific reader. Name the person and situation: freelance designers chasing overdue invoices, rather than businesses seeking efficiency.
- A concrete observation. Supply the feature, customer question, change you made or thing you learned. Give the model something to say before asking it for a hook.
- Evidence and boundaries. Include the source of any number, its timeframe and what it measures. Say when you have no measured outcome.
- Your own voice. Share a short passage you wrote and like. Explain what to preserve: plain language, a dry joke, a short opening or a personal detail.
- One next action. Decide whether the reader should try a feature, read a breakdown, answer a question or simply understand the point. A post does not always need a sales CTA.
How do you sell the result without inventing one?
Available facts: a product shows overdue invoices grouped by client, and users can send reminders from that view. No payment-speed data is available.
We’re thrilled to introduce a game-changing update that will transform the way freelancers manage their finances. Say goodbye to late payments forever.
Chasing an overdue invoice? See what each client owes and send a reminder from the same screen. Our new invoice view is live.
The revised post connects a real feature to a recognizable task. It removes the unsupported promise to eliminate late payments. It does not replace that promise with a made-up conversion rate or time saving.
When should you quantify a claim?
Use a number when you can explain where it came from and what it covers. If you measured your own process, say so. If the result came from a small customer trial, preserve the size and conditions. A personal observation should not quietly become a claim about every customer.
Without an outcome measurement, be concrete about the behavior instead: what the product shows, which step it removes or what the customer can do next. “Send reminders from the invoice view” is useful even without a percentage attached.
How do you keep an AI draft sounding like you?
Keep details that only you could have supplied: what surprised you, the tradeoff you chose or the question a customer actually asked. Ask the model to edit those sentences for clarity before asking it to replace the whole post. A smooth rewrite that removes the point of view can make a personal account feel anonymous.
Read the draft aloud and mark the phrases you would never say. Then look for repeated ideas and unnecessary setup. You do not need to remove every question, fragment or familiar expression. Those choices can be part of your normal voice.
Edit this X post for clarity. Keep my point of view and the product facts. Remove repeated ideas and unsupported hype. Do not add numbers, customer quotes or results. Use UNSLOP before you finish.
How can you tell whether the posts work?
Choose the outcome before publishing. If the goal is signups, track visits and signups from the linked post. If the goal is learning, record useful replies and the questions they raise. Likes alone do not tell you whether someone understood the offer or became a customer.
Compare posts with similar topics, audiences and offers over a consistent observation window. Record the draft, the edit and the result. Timing, distribution and the offer itself can affect performance, so one successful post does not prove that a rewrite caused the change.
UNSLOP helps with the writing step: clearer claims, less repetition and edits checked against your context. More engagement, followers and customers are outcomes to measure in your own publishing workflow, not promises a checker can make for every post.
Make the next draft clearer.
Give UNSLOP to your agent. Keep writing where you already work.
Read https://www.unslop.live/SKILL.md and set up UNSLOP for this project.