Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.agentweb.pro/llms.txt

Use this file to discover all available pages before exploring further.

This is a workflow for producing LinkedIn posts in your founder voice when you don’t want to draft from scratch. It works well for reaction-style content — your take on something someone else posted, news in your space, or a customer story you saved. The short version: save inspiration as you find it, drop it into a chat with Emma where your persona is already loaded, brain-dump your reaction, let Emma codify it into a post, then ship with screenshots and UTMs.

​
Set up once

Two things to do once so the workflow is fast every time after.

​
1. Save your persona to Brand

Emma writes in your voice when she has a Founder profile to read from. Make sure your founder voice is captured in Brand → Brand Guidelines → Founder profile. The more specific the voice description, the better. Include:
  • The way you open posts (statement, contrast, question).
  • Your sentence rhythm (short bursts, long arguments, mix).
  • Words and phrases you use a lot.
  • Things you never say (corporate filler, “exciting to announce”, em-dashes, etc.).
  • Three or four examples of posts you’ve already published in this voice. Add them to your Knowledge Base.

​
2. Set up a “LinkedIn post” scheduled chat or work plan

Save the framing prompt so you don’t retype it every time. Two options:
  • A scheduled chat with the prompt pre-loaded as the welcome message. Reuse the same chat each time.
  • A work plan with steps for the workflow below (Work plans).
The prompt itself, paraphrased: “I’m going to share a screenshot or context for a post I want to react to. Help me draft a LinkedIn post in my founder voice. Use Brand → Brand Guidelines → Founder profile for tone. Match my Knowledge Base examples for structure. Wait for me to brain-dump before you draft.” The “wait for me to brain-dump” part is important — Emma should not jump to drafting on the first message.

​
The workflow

1

Save inspiration as you scroll

When you see a LinkedIn post, X post, or article you have a reaction to, screenshot it (or save the URL). Build a small backlog so you’re not hunting for material when it’s time to post.Reaction-based content beats trying to invent original takes from a blank page. Most of your best posts will be a sharp response to something specific.
2

Open your saved post chat

Open the scheduled chat or work plan you set up above. The persona prompt is already there, so Emma is in the right mode from message one.
3

Send the source and context

Upload the screenshot of the post you’re reacting to. Or paste the text if it’s text-only. Tell Emma:
  • What the source is (e.g. “This is a post from [name] arguing X.”)
  • Why you’re reacting to it (one line — agree, disagree, complicate, extend).
Don’t ask for a draft yet.
4

Brain-dump your reaction

Just type your raw take — bullets, half-sentences, contradictions, all of it. The more honest and specific the dump, the more the post will sound like you.Things to include in the dump:
  • The specific claim or framing you’re responding to.
  • Your actual opinion (not the one you think you should have).
  • One or two pieces of evidence — a number, an example, a customer story.
  • The provocation or call you want the reader to sit with.
5

Ask Emma to codify it into a post

Now ask: “Draft this as a LinkedIn post in my founder voice.”Emma uses the screenshot + context + your dump + the persona + the KB examples to produce a draft.
6

Edit the draft

The first draft is usually 80% there. Fix the rest:
  • Tighten the opening line until it can stand alone as a hook.
  • Cut adjectives and adverbs.
  • Remove anything that sounds like a press release.
  • Make sure the post has one clear point. Two competing claims dilute both.
If a section is wrong, tell Emma what’s wrong specifically — “the second paragraph is too soft, make the claim stronger” — not just “redo it”.
7

Add a CTA if it fits

If the post genuinely connects to something AgentWeb offers, add a short CTA at the end. Examples:Don’t force a CTA on posts that don’t earn one. Audiences smell it.
8

Add UTM parameters to links

Tag every link with UTM parameters so you can attribute traffic later. Pattern:
?utm_source=linkedin&utm_medium=organic&utm_campaign=<post-topic-slug>
Ask Emma to add UTMs if she didn’t already: “Add UTM parameters for LinkedIn organic on every link in this post.”
9

Publish with the screenshot

Post on LinkedIn with the source screenshot attached. The screenshot does two jobs:
  • Shows readers what you’re reacting to without forcing them to click through.
  • Increases dwell time on the post, which LinkedIn’s algorithm rewards.
If you’re posting from inside AgentWeb via the LinkedIn integration, attach the screenshot before publishing. If you’re posting manually on LinkedIn, copy the final text out of the chat and attach the image there.

​
What makes this work

Three things, in order of importance:
  1. The brain-dump. Emma can’t fake what you actually think about something. Skip the dump and the post sounds AI-written.
  2. A specific persona description with real examples in Knowledge Base. Generic voice prompts produce generic voice.
  3. Reaction-based source material. Original takes from a blank page are hard. Reactions to specific things are easy and travel better.

​
Anti-patterns

  • Asking for “10 post ideas” with no source material. Emma can do this. The output will not sound like you and will not perform.
  • Posting the first draft unedited. Always edit. Even the best drafts have one line that gives away the AI.
  • CTAs on posts that don’t earn them. A CTA on a reaction-style post about industry news rarely fits. Save the CTA for posts where the connection is genuine.
  • Posting without a screenshot when reacting to something specific. The screenshot is the proof. Without it, readers don’t have context.

​
Variations

  • Long-form reaction: same workflow, but ask Emma to draft the post as a 250–400 word essay instead of a short hot take.
  • Carousel: dump the reaction, then ask Emma to break it into 6–8 carousel slides. Generate visuals in Creative → Images.
  • Thread on X: same workflow, but ask for a thread format instead of a single post.
  • Comment, not post: the same persona prompt works for drafting comments on someone else’s post when you’d rather engage than publish your own.

​
Next steps