The craft · AI-for-X · creatorsFor senior operators publishing under their own namePairs with /comparison/advisor-vs-influencer

AI-for-X · creators and operator-publishers

AI on the publishing line: more output, same voice, no fabrication.

Used badly, AI turns an operator with a voice into a content account that sounds like every other content account inside three months. Used well, it triples the throughput on idea capture, repurposing, and back office while leaving the voice and the source intact. This manual is the line, the workflow, and the second-brain stack a one-person publisher can run.

AudienceFounders, principals, solo creators
First deploy~one week
Hard floorVoice, source, disclosure
See the scope diagram The audience-building manual
Operator-publisher running an AI-supported content workflow.

What this work actually is

AI for creators is volume throughput on a workflow whose only durable asset is voice.

AI for creators is the deliberate use of LLMs on idea capture, drafting support, repurposing across formats, scheduling, and comment triage. It does not generate the take. It does not invent the source. It does not appear under your name without the human edit pass.

The hard floor is voice integrity, source provenance, and disclosure. Workflows that respect all three compound into a recognizable body of work. Workflows that do not produce three good months and a churned audience.

The voice is the asset. Volume without voice is the failure case, not a milestone.

The scope diagram

Three columns. The middle one is the throughput. The first one is the asset.

Human owns

What the creator does and AI does not.

  • The take and the position.
  • The source: real work, real notes, real people.
  • The headline and the closing line.
  • Reply to a sensitive comment or DM.
  • Disclosure of who is involved.
  • Any post under your name with a fact in it.
Non-delegable · the voice itself
AI assists

Where AI cuts the volume cost.

  • Idea capture from voice notes and meeting notes.
  • Outline scaffolding from your bullets.
  • Repurposing one piece into newsletter, post, thread, talk outline.
  • Editing pass for length, structure, and constraint compliance.
  • Headline candidates against your house rules.
  • Scheduling, formatting, and routine comment triage.
Drafted by AI · edited and signed by you
Never AI

What does not appear under your name.

  • Generated quotes attributed to real people.
  • Statistics and citations the model invented.
  • Positions you do not hold, generated for engagement.
  • Personal stories you did not live.
  • Replies to readers that pretend to be you.
  • Anything that breaches a named-source check or a disclosure rule.
Hard floor · audience trust
Figure 01 · scope diagram Drawn version sits here in v2: HUMAN (the voice) · AI ASSISTS (the volume) · NEVER (fabrication). Used as the back of the publishing schedule card.

What changes when this is done well

3x
Output across formats
One piece becomes newsletter, post, talk outline, video script. Same voice across.
~5h
Per week, reclaimed
From idea capture, repurposing, and back office. Day job stays the day job.
0
Generated facts shipped
Every number, name, and quote traced to a real source before publication.
1
Voice file
Single page that defines the writer the model is allowed to imitate.

What you need before you start

Five prerequisites. Three of them protect the voice; two protect the audience.

01 · A voice file

One page. How you write. What you do not.

Three sentences you would never write. Five turns of phrase you reach for. Banned vocabulary. Refusal lines. Loaded into every drafting brief.

02 · A note-capture habit

You are recording what you actually do, weekly.

The pieces draw from real work. Without the notes habit, AI fills the gap with plausible content and the voice file cannot save the piece.

03 · A no-fabrication rule

Every fact in a published piece is traceable to a real source.

Quotes, statistics, anecdotes, named people. If it is not traceable, it is cut. The rule is enforced by you, on every piece, before publication.

04 · A named-source check

Anyone identifiable would be alright with the line.

The check is mechanical: list the names a reader could identify, ask the question, anonymize or cut. Most relationship damage from publishing comes from skipping this.

05 · A disclosure pattern

How you label AI involvement.

Footer line, byline addendum, public policy page. Pick one and hold it. Audience trust survives disclosure; it does not survive being caught in something undisclosed.

The split workflow

Six publishing workflows. Each shows what you do and what AI handles.

01

Human owns

Live the work. Take the notes.

Real conversations, real meetings, real decisions. Your notebook or your voice memos are the source. Every piece traces back here.

AI assists

Idea capture from voice notes and meeting notes.

Transcribe the voice memos in your tenant. Pull the candidate themes that came up more than once. Tag them against your standing topics.

02

Human owns

Pick the take. Decide what you actually want to say.

The hardest candidate from the list. The position you are least sure of. Owned by the writer.

AI assists

Outline scaffolding from your bullets.

From three or four lines, AI proposes a structure. You keep, reorder, or scrap. The structure is yours, the scaffolding is the assist.

03

Human owns

Write the piece in your voice.

Headline, opening, the load-bearing paragraphs, the close. Always you. The voice file is the floor; you are the ceiling.

AI assists

Editing pass for constraint compliance.

Word count, banned vocabulary, sentence-shape tells. AI flags; you edit. The piece does not leave the voice.

04

Human owns

The named-source check.

Read the piece looking for anyone identifiable. Anonymize, ask, or cut. Owned by the writer, not the model.

AI assists

Surface candidates that look like real people.

Scan the draft and list every entity that reads as a real person, real company, or real situation. The writer decides what each is.

05

Human owns

Publish to your owned surface first.

Newsletter, blog. Then platforms. The owned surface protects you when an algorithm changes.

AI assists

Repurpose into newsletter, post, thread, talk outline.

One piece, multiple surfaces. The voice file holds across formats. You sign each off.

06

Human owns

Reply to readers in real ways.

Your name is on the reply. AI does not pretend to be you in DMs, comments, or email.

AI assists

Comment triage and routine response drafts.

Sort comments into categories: ignore, like, reply. Draft routine responses for batched human review. The writer sends.

Figure 02 · publishing swim lanes Six-step swim-lane diagram lives here in v2. Top lane: writer. Bottom lane: AI. Voice file runs across all six steps as the constant constraint.

How to know AI is hurting the voice

Six tells the throughput has flattened the writer.

Tell 01

Two long-time readers said it sounds different.

The voice slipped. Pull the last six pieces, read them next to your older work, rebuild the voice file from the older ones.

Tell 02

A piece quoted someone who never said that.

Fabricated quote. Repair the piece, post a correction, add the failure to the no-fabrication rule, run the named-source check on the next three drafts.

Tell 03

A statistic in a published piece cannot be traced.

Hallucinated source. Pull the line, repair, sharpen the no-fabrication rule, treat as a near-miss in the brain.

Tell 04

You took a position you do not actually hold to fit a format.

The format is now writing the piece. Reverse the flow: write the take first, then choose the format.

Tell 05

A real person flagged a piece they recognized themselves in.

The named-source check failed. Repair the relationship, repair the piece, tighten the rule.

Tell 06

You publish more than you do real work.

The publishing has consumed the operating week. Cut the cadence in half. The audience that compounds prefers fewer, sharper pieces.

Tools and tactics

A second brain shaped for a one-person publishing line.

The brain holds the voice file, the candidate list, the named-source log, and the published archive. The voice is the asset; the brain protects it.

The Second Brain · publishing edition

Stan's adapted stack for creators

Voice file pinned at the top. Candidate list refreshed weekly from notes. Five named brief blocks: idea capture, outline scaffold, edit pass, named-source check, repurpose. Published archive with source notes, what got cut, audience response. Quarterly review reads three months blind and asks whether the voice held.

  • Voice file pinned, refreshed monthly.
  • Candidate list, refreshed weekly.
  • Five named brief blocks.
  • Published archive: source, draft, cuts, response.
  • Quarterly blind read of three months of pieces.

Documented in full inside the engagement · teaser here

Tactic 02

The voice file

One page. Three sentences you would never write. Five turns of phrase you reach for. Banned vocabulary list (AI tells, generic creator phrases). Refusal rules. Loaded as the system prompt for every drafting brief.

  • One page, plain text.
  • Refreshed monthly.
  • Versioned and dated.

Tactic 03

The no-fabrication rule

Standing rule before any publication: every fact in the piece is traceable to a real source. Names, numbers, quotes, dates. The rule is enforced by the writer, on every piece, before the publish button.

  • Rule pinned to the publishing template.
  • Checklist at the bottom of every draft.
  • Failures recorded in the brain so they do not repeat.

Tactic 04

The disclosure footer

One line at the bottom of every relevant piece. Names what AI touched: drafting support, editing pass, repurposing. Audience trust survives disclosure; it does not survive being caught in something undisclosed.

  • Same line, every piece.
  • Specific workflow names.
  • Linked to a public policy page.

Coming soon

Three rooms held open inside this manual.

Built for operator-publishers who want the stack pre-assembled.

In build

The Operator-Publisher Stack

Voice file template, candidate list, five named briefs, named-source checklist, disclosure footer, repurpose templates. Released after one year of unchanged form.

Scoped

The Voice Audit

A small structured engagement: read of the last six pieces, voice diagnosis, voice file rebuild, weekly review reset.

Scoped

The Repurpose Library

Templates for turning one piece into newsletter, post, thread, talk, video script without losing voice.

What this work is not

An audience earns trust. AI does not earn it back once it is lost.

This page makes a publisher faster, not louder.

The throughput is on volume. The asset is the voice and the trust. The comparison page sets the structural difference between the take that scales and the read that closes a decision.

Read advisor vs. influencer →
Pull AI back out of the workflow when
  • The voice has slipped and a long-time reader noticed.
  • A fact in a published piece cannot be traced.
  • You took a position you do not hold.
  • A real person flagged a piece they recognized.

When the audience question is the wrong question

Run the stack for one quarter.
If the audience grew and the business did not, the strategy is upside-down.

Application-gated. Personal reply within 48 hours.

Apply for advisory

Tier 01 from $2,500 · Tier 02 from $4,500 / month · All three tiers