Twice-weekly LinkedIn thought-leader posts with matching AI illustrations

Every Tuesday and Thursday at 8:20am in the operator's local timezone, publish a thought-leadership post to LinkedIn on my behalf, paired with a custom AI-generated illustration so the post stops the scroll. Trigger is cron only.

Inputs the operator will supply when configuring the workflow: (1) two or three content pillars they want to be known for, with one or two sentences of context per pillar; (2) a short voice guide (tone, sentence rhythm, phrases they do and do not use, whether they open with hooks or stories); (3) a brand visual style block (palette, mood, aesthetic tokens like flat editorial, isometric, muted editorial gradients, etc.); (4) the Slack channel or DM that should receive the preview; (5) optionally, a brand mark or reference image URL to weave into the illustrations.

Step 1 — pick a fresh angle. Rotate across the content pillars and vary the angle. Do not repeat the topic or framing of the last few posts. If the operator's recent post history is available, skim it and pick something adjacent, not identical.

Step 2 — draft the post. Write a 900 to 1400 character LinkedIn post in the operator's voice. Open with a strong hook, use short paragraphs and single-line breaks the way LinkedIn renders best, and close with a light question or observation to invite comments. No hashtag walls, at most two or three relevant hashtags at the end. Never use em dashes.

Step 3 — generate the illustration with fal.ai. Use Generate Image with FLUX1.1 [pro] by default at 1200x1200 for editorial illustrations. Switch to Generate or Edit Image with Nano Banana (Gemini) only when the operator has supplied a brand mark or reference image URL that must be woven into the visual. Every fal.ai prompt must include the brand style tokens from the operator's visual style block (palette, mood, aesthetic like flat editorial or isometric) so illustrations stay visually consistent post to post. fal.ai output images are hosted on v3.fal.media and can be referenced directly by URL when creating the LinkedIn image post.

Step 4 — send the pre-publish Slack preview. Use Send a Message on Slack to drop the drafted commentary and the fal.ai image URL into the operator's chosen channel or DM, with a note like ‘Publishing to LinkedIn at 8:30am unless you reply veto in the next 10 minutes.’ This is the operator's chance to kill it.

Step 5 — wait roughly ten minutes, then check for a veto reply in the same Slack channel. If any recent reply from the operator contains a veto word (‘veto’, ‘stop’, ‘hold’, ‘no’, ‘skip’), do NOT publish; instead post a short confirmation in Slack that the post was held, and stop.

Step 6 — publish to LinkedIn. Use Create Post on LinkedIn as an image post, with the drafted commentary as the post text and the fal.ai image URL as the attached image. After posting, drop the LinkedIn post URL back into the same Slack thread so the operator has one place to see everything that shipped.

Nuance to bake in: keep the illustrations visually consistent by ALWAYS injecting the brand style tokens into the fal.ai prompt (palette, mood, ‘flat editorial’, ‘isometric’, whatever the operator specified). Skipping that turns the feed into a random art gallery. Also: image posts on LinkedIn consistently outperform text-only, and the reason most people do not do this every time is drafting a fresh visual — fal.ai closes that loop.