How Sonny (Marblism's AI Employee) Built a Month of LinkedIn Posts in One Conversation

Trying to plan LinkedIn content solo usually ends in "I'll post more consistently next month" — and nothing happens. This video shows what changes when you ask an AI employee: instead of a template, Sonny (Marblism's AI marketer) asked clarifying questions that produced a full day-by-day, hour-by-hour LinkedIn schedule with themes and post ideas, plus a new way to think about content.

The Before State

The creator's honest diagnosis of DIY LinkedIn planning:

Zero structure — no rhythm to how posts get made
No schedule — ad-hoc posting whenever inspiration strikes
No themes — every post is a separate decision with no connective tissue
No clue what to post — just guessing, hoping something sticks

This is the state most founders and operators are in with social. It's not a content problem. It's a system problem.

What Sonny Did Differently

The creator asked: "Can you help me scale my LinkedIn posts for next month?"

Sonny's response wasn't a template. It was clarifying questions:

Question	Why it matters
How many posts per week?	Calibrates the total volume and effort
What main themes do you want to cover?	Creates narrative consistency across posts
Which days and times work best for your audience?	Schedules around when the audience is active

Three questions the creator hadn't even thought to ask themselves.

The Output

By the end of the conversation, the creator had:

Full posting schedule — day by day, hour by hour
Clear themes — so each post reinforces a consistent narrative
Specific post ideas — not placeholders, actual angles to write
Confidence — the emotional output, which matters more than most people admit

The Real Unlock

The creator names it explicitly:

"Sonny didn't just help me create posts. He taught me how to think about content in a way I never had before."

That's the gap between AI as tool and AI as employee. A tool takes your prompt and produces output. An employee asks the questions you should have asked yourself. The asking is the teaching.

Why This Works — The Diagnostic Prompt Pattern

When an AI asks questions before producing output, three things happen:

1. The problem gets scoped

You realize "plan LinkedIn posts" is actually five different problems (cadence, themes, timing, formats, distribution). The questions surface the scope.

2. You do the thinking

The AI doesn't invent your audience or your themes — you articulate them in response to the questions. That articulation is the strategy.

3. The output reflects your actual constraints

Because the AI's output is conditioned on your answers, the schedule works for your audience, your bandwidth, your goals — not a generic template.

What "AI Employee" Means in Practice

Marblism's framing is worth unpacking: AI employees differ from AI assistants in three ways:

AI Assistant	AI Employee
Reacts to your prompt	Proactively drives the conversation
One-shot outputs	Multi-turn conversations that build context
Generic default output	Role-specific expertise (marketing, sales, support)

Sonny is the marketing version. The implication is there are others — sales, support, ops — each playing the same pattern: asks the questions, generates the output, keeps context across interactions.

How to Get This Without Marblism

If you want the same workflow with plain Claude or ChatGPT:

Install a content strategy skill (the 29-skill Corey Haynes library works)
Prompt: "Before giving me anything, ask me clarifying questions about my audience, goals, and constraints."
Actually answer the questions — don't short-cut them
Iterate — treat the output as a draft schedule, refine with follow-ups

The Marblism version is nicer UX; the underlying pattern works anywhere.

Key Takeaways

DIY LinkedIn planning fails because it's a system problem, not a content problem
Sonny (Marblism AI marketer) scaled a month of LinkedIn content by asking three clarifying questions: post frequency, themes, ideal days/times
The output: a day-by-day, hour-by-hour posting schedule with themes and post ideas
The real unlock: Sonny taught the creator how to think about content — asking was the teaching
Replicable pattern: prompt AI to ask diagnostic questions before producing output

Resources

Marblism — Source of Sonny, the AI employee platform
Claude — Use with skills for the DIY version of this workflow
ChatGPT — Works with custom GPTs for a similar diagnostic pattern