AI generated social media posts save your team the blank-page hour, not the thinking. Used well, they turn one approved idea into a week of drafts in the voice you have already set, ready for a person to check and schedule. Used badly, they fill your feeds with posts that could belong to any firm in your sector. This article explains how to create social media posts with AI inside the levelpartner.ai platform's social planner, what to feed the AI so the drafts sound like you, and where a human still has to sign off.
What AI drafting is actually for
A 60-person firm rarely lacks things to say. It lacks someone with a free afternoon to turn a case note, a webinar or a partner's opinion into short posts for four networks. AI drafting closes that gap. You give it the raw material and the angle, it returns a caption per network, and your marketing lead edits rather than writes.
That shift matters most when one person carries social alongside other work. Editing five drafts takes a fraction of the time of writing five posts. Before any draft is scheduled, the reviewer reads it in full. The quality of the result depends almost entirely on the brief you give and on who reviews the output before it goes live.
Before you start
- Your social accounts connected in the Social Planner, with posting permission confirmed for each one.
- A short written voice guide: who you write for, words you use, words you never use, and two or three posts you consider good.
- A list of topics you are happy to post about and a shorter list of topics you never touch, such as rates advice or client names.
- One named reviewer who approves AI drafts before they are scheduled.
- A Brand Voice saved under Marketing > Brand Boards, if you want drafts to follow it.
- A usage budget agreed, since AI content is billed by the word as it is used.
How to set it up
- Go to Marketing > Social Planner.
- Click + New Post and choose to create with Content AI.
- Select the social platform and the content type.
- Select your Brand Voice if you have saved one.
- Answer the required Guided Content Questions with your source material.
- Click Generate.
- Review the draft for accuracy, tone, links and calls to action.
- Edit manually, or use the AI-assisted editing options.
- Publish now or schedule the post.
Feeding the AI so drafts sound like your firm
Generic input produces generic output. If you ask for "a post about mortgage rates", you get a post any broker could publish. If you paste three sentences from a real client conversation, stripped of identifying detail, and ask for a post on the one question that client asked, you get something with an edge.
Keep a running document of raw material your team already produces: answers to good client questions, short observations from sales calls, notes from a partner's talk. That document becomes the input queue for AI drafting. It also stops the AI from inventing facts, because every draft starts from something true.
Suppose the consultancy runs a 45-minute webinar each month. The marketing lead could pull six points from the transcript and ask for a LinkedIn post and a shorter version for X on each: twelve drafts. If editing kept nine of them, that would be roughly two approved posts a week for four weeks, all from one hour of material the firm had already produced.
What AI drafting replaces, and what it does not
It replaces the first draft and the reformatting work: cutting a long caption down for one network, rewording the same point so it does not appear twice in a week, proposing three hooks to choose from. It also helps with image ideas if you use AI image generation, though for most professional services firms your own photography carries more trust.
It does not replace judgement. The AI does not know which client relationship is sensitive, which regulator is watching your sector, or that your managing director dislikes a particular phrase. Mortgage brokers and real estate firms in particular need a person to check anything that touches on rates, guarantees or property claims. If you are planning the wider month, see how to plan social media posts for the routine that sits around the drafting.
Weekly time it takes
Budget around ninety minutes a week once the voice guide is settled: thirty minutes collecting raw material, thirty minutes generating and editing, and thirty minutes for review and scheduling. In the first month expect double that while you learn which prompts work. Put the approved drafts straight onto your social content calendar so nobody has to remember where they were saved.
The most common mistake is publishing the first draft. AI copy tends to open with a broad statement and close with a question nobody will answer. Rewrite the first line and cut the closing question, and the post usually improves more than any amount of prompt tuning would achieve.
When not to use it
Do not use AI drafting for anything announcing a client result, a personnel change or a response to a complaint. Those posts need a person to write them from the start. Avoid it too when the source material is thin. If you have nothing specific to say, a well-written empty post is still empty, and it trains your audience to scroll past you.
Troubleshooting
Every draft sounds the same. Your prompt is doing the same job each time. Change the angle line, and paste different source material rather than asking for variations of one idea.
Drafts include claims you cannot support. The AI filled a gap in your input. Give it the facts you want used and tell it not to add statistics or results. Then check each draft against that instruction.
Captions are too long for some networks. Generate one master caption, then use the per-network edit to shorten it. The guide to scheduling a single post covers per-network customisation in detail.
Usage is higher than expected. Someone is regenerating drafts many times instead of editing. Set a rule of two generations per post, then edit by hand.
Frequently asked questions
Will people know a post was written with AI?
They will if it reads like a template. Posts that start from your own material and are edited by someone who knows your clients are hard to tell apart from posts written by hand. Readers care more about whether the post is useful.
Can AI write posts for all my networks at once?
Yes. You can generate one draft and adapt it per network in the same post editor. Shorten and adjust each version rather than publishing the same caption everywhere, since what works on LinkedIn often reads badly elsewhere.
Is it safe for a regulated firm to use AI drafts?
It is as safe as your review step. Keep a named approver, keep a list of topics the AI must not cover, and never schedule a draft that has not been read by that person. For broader controls across channels, see manage social media accounts.
Does AI drafting cost extra?
AI content is billed as used on usage, separate from the licence. A disciplined team that edits rather than regenerates keeps that usage low.