Generic prompts produce generic social posts, and the good ones you do write get stranded in a chat window you cannot find again. Loren Bartley fixed both with the same move: over 150 named skills inside one Claude project, handing work to each other until a week of posts lands in Agorapulse as drafts.

KEY TAKEAWAYS

A skill carries your standards as well as your instructions, so the work runs to the same bar every time instead of drifting with whatever you typed that day.

A chat with no memory of your past positions will contradict you across a month of posts, and you will miss it because you did not write them.

Loren rewrites her old talks with a skill called Trini before they enter her knowledge folder, so her ums and tangents never become source material.

Constance plans the quarter, Edison plans the week, Connie assigns the batch, and each one saves its output for the next cycle to read.

Naming 150 skills after their jobs turns the list into a retrieval system, which is how Audrey Ads Auditor gets found.

Agatha checks the approved rows before staging them and caught Loren attaching a still image to a post briefed as video.

CHAPTERS

00:00 Why another post stopped feeling like a chore

00:52 The two bottlenecks that had nothing to do with writing

02:17 Why a fresh chat every time produces contradictions

03:41 What a skill holds that a saved prompt does not

06:22 Inside the Content Accelerator project and its knowledge files

08:12 Trini rewrites her old talks before they enter the knowledge base

09:29 Constance builds the quarterly content reference guide

11:38 Edison turns the strategy into a weekly cycle brief

15:09 Naming 150 skills so she can find the right one

16:22 Connie assigns the batch to Caprice, Vivian and Hugo

17:51 Hubert's Notion hub replaces posts buried in chats

21:13 Agatha stages the approved rows into Agorapulse

23:17 Agatha catches the still image on a video post

26:22 What Loren does with the time she got back

28:04 Where to get Edison

Loren shares her screen for most of this session, so the video is worth a look on YouTube.

Full episode breakdown and the blueprint: https://www.agorapulse.com/blog/ai-social-playbook/e6/

Loren is giving away Edison, the skill that builds her weekly cycle brief: https://aimegasuccess.com/academy

CONNECT WITH LOREN BARTLEY

Website: https://impactiv8.com.au/

LinkedIn: https://www.linkedin.com/in/lorenbartley/

Instagram: https://www.instagram.com/impactiv8/

THE AI SOCIAL PLAYBOOK

YouTube: https://www.youtube.com/@TheAISocialPlaybook

Apple: https://podcasts.apple.com/us/podcast/the-ai-social-playbook/id6808658318

Spotify: https://open.spotify.com/show/4tmvz9PRDHA9qX2SCoFF8h

Connect with Mike Allton: https://www.linkedin.com/in/mikeallton/

Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Most social content gets built from random topics or from pillars that name subjects rather than objections, which earns engagement and no sales. Molly Mahoney maps the seven beliefs a buyer has to accept first, then turns one belief into a week of posts.

The seven belief categories are why me, why this, why you, why now, why not, why stay, and why worse.

Why stay is the category subscription products and memberships depend on, because the customer decides again every month.

Molly typed 9/11 into the why-now prompt on purpose and got a holiday-overwhelm line she called tacky and disrespectful. A human still decides which dates carry weight.

Her earlier version wrote a month of content at once and none of it got used, so this one caps the output at five posts from one belief.

A finished batch exports as a CSV and merges with the Agorapulse bulk upload template, so a week queues without retyping.

One client, a Toastmasters speaker who had only sold to friends and family, sold a $6,000 AI assessment and setup to a stranger after publishing against these beliefs.

CHAPTERS

00:00 The snarky meme that made Mike wince

00:38 Why content pillars don't move buyers

03:52 Brand first, beliefs second, one week of content per belief

07:29 One click, four beliefs per category

09:37 The 9/11 belief AI didn't know to refuse

10:11 Why stay, the category memberships and SaaS live on

12:06 Inside the story bank: quotes, metaphors, memes

14:58 Why the tool refuses to write a month at once

15:39 The five-post batch, conversation starter to POV meme

19:18 CSV export straight into Agorapulse bulk upload

22:55 The $6,000 sale to a total stranger

https://www.agorapulse.com/blog/ai-social-playbook/e5/

CONNECT WITH MOLLY MAHONEY

Website: https://mollymahoney.com/

LinkedIn: https://www.linkedin.com/in/mollygolive

Instagram: https://www.instagram.com/mollymahoneyofficial

Connect with Mike Allton: https://www.linkedin.com/in/mikeallton/

Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Most social teams brainstorm content ideas from memory or a generic AI prompt, then wonder why nothing sounds different from a competitor's post. Christopher Penn, Chief Data Scientist at TrustInsights.ai, pulls real audience questions straight out of Reddit and turns them into a ranked list of 100 video topics.

In this episode:

• A model on par with Claude Sonnet 5 can run for about 1/40th the cost through a smart-and-cheap provider pick.

• Reddit's questions, not its answers, make the safest source material, with a human still filtering trolls and skipping competitor mentions.

• One Gemini Notebook prompt turns a raw Reddit export into 100 ranked questions.

• Recording on an iPhone in a kitchen can beat a polished studio setup on platforms downranking synthetic-looking video.

• Turning on YouTube's third-party training permission gets a creator's expertise cited by AI agents.

Timestamps:

0:00 - The line that opens every failed AI pipeline

2:39 - Why Chris skipped Claude Code for OpenCode

5:05 - Mapping models by smart-and-cheap, not by brand

9:45 - Why a product requirements document comes before any code

13:31 - The one file that survives an AI's memory getting compressed

17:50 - Turning a Reddit export into a grounded notebook

20:51 - The exact prompt that produces 100 ranked questions

22:05 - Why a human still filters trolls and skips competitors

25:00 - Recording in a kitchen on purpose

29:07 - The YouTube setting that trains AI on your expertise

31:52 - The one trap that breaks every rushed AI pipeline

Guest: Christopher Penn, Chief Data Scientist, TrustInsights.ai

trustinsights.ai · christopherspenn.com · LinkedIn

Full show notes and transcript:

https://www.agorapulse.com/blog/podcasts/ai-social-playbook/e4

Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.