30-second summary
What it is, in three sentences.
The IEXDG Brain reads your strategy doc, picks a real Shutterstock photo, and drops three draft posts (LinkedIn, Facebook, Instagram) into your GHL Social Planner. You review on your phone or laptop, edit anything that feels off, then click publish. Nothing is ever published automatically, and five hard gates check every draft before it leaves the Brain.
What it actually does, in plain English
No code. No abstractions. Here is the day, end to end.
The daily rhythm
- The Brain pulls from your doc. Each morning the Brain reads the canonical strategy doc for the day (which pillar, which hero, which sub-vignette, which signature hook).
- It picks a real photo. Shutterstock-primary search runs against the pillar's emotional bullets. The image passes the visual gate (cream and navy palette, no staged stock, no fake leaders) or it gets thrown out and the Brain tries again.
- It writes three separate drafts. LinkedIn long-form, Facebook conversational, Instagram carousel. Each is its own draft, not one combined post copy-pasted to three places.
- It drops them into GHL Social Planner as DRAFTS. Rule 12 is absolute. The Brain never publishes. Status flips to scheduled or published only when you click it.
- You review on your phone. The HighLevel app gives you the native review path. The Brain also exposes a mobile-friendly tile at
brain.iexdg.com/drafts.htmlwhen you prefer the dashboard view. - You publish, the first comment fires. Within 60 seconds of publish, the saved first-comment artifact gets pasted (engagement driver, authority expansion, CTA, or resource drop, picked by post type).
The 5 hard gates that fire before every push
Every draft passes through every gate. Any failure stops the push and writes to the audit log. Nothing leaks to GHL until all five are green.
Metadata leak guard
Scans the body for any internal scaffolding token (pillar markers, sub-vignette tags, gate scores, run IDs). If anything from the Brain's own machinery shows up in the caption, the push is killed. Captions read like a human wrote them, not like a system spit them out.
Em-dash audit
Em dashes are forbidden across every IEXDG artifact. The audit walks the caption and the first comment, character by character, looking for any em dash. The only allowed em dash is inside her verbatim attribution line. Everything else fails the push.
Watermark and asset guard
Every image is sanity-checked for filename, file size, and dimensions before upload. Watermarked stock or low-resolution junk gets flagged and the Brain re-picks. Real photo only, professional quality, no exceptions.
DRAFT-only enforcement
Every payload that leaves the Brain has its status explicitly set to draft. The push code refuses to submit anything else. You see it before any audience does. Always.
Asset separation
Body, first comment, and carousel slides stay separate. They never get merged into one blob. Each surface gets its own asset, its own role, its own gate. The Brain treats them as different deliverables that ship together, not the same thing copy-pasted.
What happens when a gate fails?
The push aborts. The run log under source/_audit/run_logs/ records which gate failed, what the offending text was, what the Brain tried next. The Brain either re-rolls (new image, new caption variant) or escalates to Robert for manual review. Nothing silent. Nothing partial.
3 separate deliverables per day
LinkedIn, Facebook, and Instagram each ship as their own draft. Not one combined post. Here is why.
Long-form, executive voice. 1,300 to 2,000 characters. One signature hook, one frame, one CTA. Built for the CPO, the Superintendent, the Director scrolling between meetings.
Conversational frame. 600 to 1,100 characters. Slight tone shift, more story, less formal. Built for the wider leader audience that lives on Facebook.
Instagram (carousel)
Carousel of 5 to 9 slides. Each slide is its own asset with its own caption fragment. Built for the leaders who live in the visual feed.
Why separate, not one combined post
Each channel has its own audience, its own attention span, its own format expectation. Pasting the same copy to three places looks lazy and underperforms on every metric. Three drafts means three reviews, but it also means three messages tuned to where they live.
The first-comment artifact
Fires 30 to 60 seconds after publish. Saved as a file. Pasted by hand or by automation.
How the artifact gets made
For every draft, the Brain also writes a first-comment file to disk under the day's content folder. The file names the variant (engagement driver, authority expansion, CTA, or resource drop), the post it belongs to, and the exact text to paste. When you publish, the first comment is one paste away.
The 4 variants, picked by post type
- Engagement Driver: "What part of this resonated with you the most?"
- Authority Expansion: "One thing I did not include in the post."
- Call to Action: "Start here: iexdg.com/culture-pulse"
- Resource Drop: "Comment DIAGNOSTIC and I will send it"
Why 30 to 60 seconds matters
The first comment in that window primes the algorithm. Comments after the first hour rarely move the post. The 60-second window is the lever.
On your phone
Two review paths. Pick the one that fits the moment.
HighLevel mobile app (native)
The HighLevel app gives you the full Social Planner inside your pocket. Edit caption, swap image, schedule, publish. Same UI as the desktop.
iOS App Store Google PlayTo edit: open app, tap Marketing, tap Social Planner, tap the draft, tap Edit, change, tap Save or Publish.
brain.iexdg.com/drafts.html (Brain tile)
Mobile-friendly tile on the Brain. Headline, image thumb, platform, scheduled date, Edit-in-GHL link. Pull to refresh. Add-a-note posts a private comment to the Brain ledger.
Open the drafts tileDirect link to the GHL Social Planner in browser: open Social Planner.
The audit trail
Every push leaves a run log. Honest, machine-readable, replayable.
Where it lives
Each push writes a JSON run log to source/_audit/run_logs/YYYY-MM-DD_HHMMSS_ghl_push.json on the VM. The log records every gate result, every payload, every response from GHL, and every retry. If a draft ever looks wrong in the Social Planner, the run log tells you exactly why.
What the log contains
- Timestamp, day_verbatim source path, channel, account ID
- Each of the 5 gate results (pass or fail, with offending text if fail)
- Image picked (Shutterstock asset ID, license type, dimensions)
- Final body, first comment, carousel slide manifests
- GHL response payload (post ID, status, scheduled date)
- Total elapsed time and any rate-limit pauses
The flow, end to end
Topic-native diagram. No transplants from other docs.