I grow my own accounts and handle brand collaborations every day with Claude and Postey. This guide is the version of AI marketing I actually use: what to hand to AI, what to keep for yourself, and a one-week setup you can copy.
What AI marketing means for a one-person business
Large companies use AI for ad bidding, personalisation and forecasting across millions of customers. A solo founder has a different problem: too many marketing jobs and not enough hours. Here is how the work splits.
| Marketing job | What AI does well | What you still decide |
|---|---|---|
| Research | Summarise customer emails, reviews and replies into themes | Which problem is worth talking about |
| Planning | Turn one idea into a week of post ideas | What you are selling this month |
| Writing | First drafts, and a version for each network | Every claim, number and promise |
| Distribution | Format, schedule and publish at set times | Which channels deserve your time |
| Replies | Draft answers to comments and DMs | Anything sensitive, angry or legal |
| Reporting | Pull the numbers and spot the outliers | What to change next week |
The pattern: AI handles volume and formatting; you own judgement and truth. Every AI marketing mistake I see comes from flipping that.
A worked example, start to finish
Here is one full cycle for a solo SaaS founder. It's an illustrative example, not a real customer.
- The input. A customer emails: “Can I send one invoice to two people at the same client?” Your product added CC contacts last week.
- The prompt. You paste the email and the changelog line into Claude or ChatGPT, with the prompt further down this page.
- The draft. It suggests a LinkedIn post: “A customer asked how to send one invoice to both their client’s finance team and their manager. Since last week, you can add both as CC contacts, and each gets the same invoice. Two clicks, no duplicates.”
- Your edit. You cut “Two clicks” because you haven’t timed it, and add a real screenshot.
- Scheduled and checked. It goes out on Tuesday at noon. On Sunday you check whether anyone replied, asked a question or booked a demo, not just how many people saw it.
The same loop works for a shop. A florist pastes a customer’s message, “Do you deliver on Sundays?”, and asks: “Write an Instagram post answering this, in a friendly tone, using only these delivery details: [paste them].” For a creator, a follower’s DM becomes a how-to post.
How to use AI for marketing: a one-week setup
You don't need a marketing stack to begin. Your strategy is three answers AI can't give you: who you want to reach, what you want them to do this month, and where they already spend time. Turn those into three things:
- One channel. Pick where your customers already are. For a developer tool that might be X or LinkedIn; for a café, Instagram; for a consultant, LinkedIn.
- One offer. The thing you want people to try, buy or book this month.
- One source of real material. Customer emails, support questions, sales calls, your changelog, or what you learned this week.
Then run this for one week, in three short sittings:
- Start of the week (about an hour): paste your material into Claude or ChatGPT and ask for three post ideas tied to your offer. Turn each idea into a draft, edit all three until they sound like you, and schedule them across the week.
- During the week: reply to comments yourself, using AI drafts only as a starting point.
- End of the week (15 minutes): look at what started conversations, not only what got views.
What that looks like in practice:
- SaaS founder: a customer's support question becomes a how-to post, a demo clip and a "how do you handle this today?" question.
- Café or shop: "do you do gluten-free?" becomes a post about the menu, a photo of the morning bake and a poll for next week's special.
- Freelancer or consultant: a client problem you solved becomes a short case study (with permission), a lesson and an invitation to book a call.
AI social media marketing, step by step
Social media is where AI saves the most time for a small business, because every network wants its own version of the same idea. This is the workflow I use with Postey.
1. Connect your AI to your accounts
Postey connects ChatGPT and Claude to X, LinkedIn, Instagram, TikTok, YouTube, Threads, Bluesky, Facebook and Pinterest through one link. Follow Post from ChatGPT or the Claude guide. The AI connection is on paid plans, and the free trial includes it.
2. Give it real material, not a topic
"Write a post about productivity" gets you the same post as everyone else. Paste the real thing: a customer email, a menu change, a brand brief, a changelog, or a follower's DM. This prompt keeps you in control:
Here is my material: [paste it]. Draft three posts for [your channel] that answer the question or share the news in it. Use only facts from my material. Don't invent numbers, quotes or results.
If your assistant is connected to Postey, add: "Save them as drafts in Postey and show me each one before anything is scheduled."
3. Review every draft
Read each version as your customer would. Cut anything you can't prove. Check that each network's version reads naturally there, not like the same text trimmed to fit.
4. Schedule, then step away
Ask your assistant to schedule the approved drafts, or schedule them in the Postey calendar yourself. By default, your AI shows you each post and asks before it schedules or publishes. If you'd rather it could never schedule on its own, connect it with drafting-only access.
This demo shows the Claude connection with a LinkedIn post; the same steps work for every network.
5. Measure what matters
Postey Analytics shows views, engagement and a best-time-to-post grid for your own posts. Look at your typical post, not your best one. When I analysed 116 of my own Threads posts, 12 breakout posts produced 87% of all views, and no posting hour reliably won. The full breakdown is in my best time to post on Threads guide.
My own routine, as one example
I'm a creator, so my routine is built around brand collaborations. Yours will look different, but the order is the same.
- Briefs come in by email. Brands email me briefs. I ask Claude to read the whole email and pull out the brief and the requirements.
- Claude drafts both sides. From that, it drafts my reply to the brand and the post itself.
- Everything goes through me. I approve the text, then make the demo videos and images myself.
- It goes back to the brand, and Postey tracks it. My inbox is connected to Claude through its own connector, so I ask Claude to email the approved work back to the brand for sign-off. Postey handles the social side. Once it's live, Postey tracks engagement and sends me alerts, so I can see which posts work and which don't.
- I build an email list, not just followers. Relying on one platform is risky, so I use Instagram comment-to-DM automation in Postey (included from the Creator plan): when someone comments a keyword, they get a DM with my newsletter link.
I prepare posts 10–15 days ahead and spend three to four hours a week on it. That time is my own estimate, not a measured benchmark, and your first weeks will take longer while you set things up.
What to measure (and what to ignore)
AI makes it easy to produce more. More is not the goal.
| Measure this | Because |
|---|---|
| Replies, DMs and enquiries from the right people | They're the closest signal to a customer |
| Clicks to your site, sign-ups and sales | Views don't pay your bills |
| Your typical post's reach (the middle result, not the best) | One viral post hides a weak routine |
| Time you spend per week | The routine has to survive busy weeks |
Ignore follower counts in the first month, and don't judge a post in its first hour. If three weeks pass with views but no conversations, change the topic or the offer, not the posting volume.
What AI shouldn't do in your marketing
Some of these are good practice; some are the law.
- Never invent reviews, testimonials or customer quotes. In the US, the FTC's rule on fake reviews, in force since October 2024, explicitly covers AI-generated reviews and testimonials, and it allows civil penalties.
- Never publish numbers you can't source. AI will happily write "used by 10,000 founders". If it isn't true, it's a liability.
- Don't mass-produce web pages. Google says it judges content by quality, not by whether AI helped. But producing lots of pages without real review is treated as spam.
- Don't automate engagement. Auto-liking, auto-following and bot comments break most networks' rules and look exactly like what they are. Comment-to-DM is different: it only messages people who asked for something by commenting your keyword.
- Keep your voice. Your customers chose you, not a model. Edit until the draft sounds like something you'd say out loud.
The AI marketing tools a small business actually needs
You don't need ten subscriptions. Most solo founders need four things:
| Job | What you need | Notes |
|---|---|---|
| Thinking and drafting | An AI assistant (ChatGPT or Claude) | A free account is enough to start |
| Publishing and scheduling | A scheduler your AI can use | One place for every network, with a calendar |
| Measuring | Analytics for your posts and your site | Posts and sign-ups, not just likes |
| Owning your audience | An email list or newsletter | Platforms change; your list stays yours |
Postey covers the second and third rows and helps with the fourth through comment-to-DM automation. It works with the assistant you already use instead of replacing it.
Postey's limits, so you can decide (prices checked 16 November 2026):
- Postey doesn't run paid ads or send email newsletters.
- On X, Postey adds a small usage charge per post ($0.02, or $0.25 for a post with a link), on top of the plan price.
- The Free plan publishes from the app but doesn't include the AI connection; that comes with paid plans and the trial.
- Creator covers one brand. Managing several brands or clients? The Brand plan is US$29/month for 5 brands, and Agency covers 20; see pricing.
Try it on your own week
Try it free and connect Claude or ChatGPT, draft a week of posts and schedule them. Creator is US$9/month after the trial. A card is required, and you can cancel before the trial ends. See plan details.
For setup detail, see Claude Code for social media or what an AI social media manager can and can't do yet.
Sources and scope
- Shruti Mishra's own workflow and Threads data: her daily routine as described 16 November 2026; 116 of her Threads posts with stored metrics, analysed 16 November 2026. Time estimates are her own, not measured.
- Google Search Central, "Google Search's guidance about AI-generated content" (February 2023) and its guidance on using generative AI content, checked 16 November 2026.
- US Federal Trade Commission, final rule on consumer reviews and testimonials: announced 14 August 2024, effective 21 October 2024, checked 16 November 2026.
- Postey plans, AI connection, analytics and X usage rates: checked in the live app and plan catalog, 16 November 2026. Pricing.
This guide covers AI for organic marketing by a small team. It isn't legal advice, and it doesn't cover paid-ads automation.



