From Prompt to Profit June 24 2025 Flowgrammer Event Recap
A huge thank you to everyone who came out to last night’s Flowgrammer AI & Business Automation Meetup! Your enthusiasm, questions, and insights made the evening a great success. We hope you w
A huge thank you to everyone who came out to last night’s Flowgrammer AI & Business Automation Meetup! Your enthusiasm, questions, and insights made the evening a great success. We hope you walked away inspired to build smarter, more efficient workflows.
🚀 Sponsor Spotlight: Floqer
We’re thrilled to have Floqer powering our GTM workflows. As a Flowgrammer attendee, you get:
- Lead enrichment from 80 + data providers
- Real-time buyer signals (funding, team growth, tech adoption)
- One-click personalized emails
- Automatic sync to your CRM or outreach tools
Floqer is hiring in Toronto! Interested in Product Design, Engineering, or Marketing? Reach out to Shivam at Shivam@floqer.com or connect on LinkedIn.
📅 Save the Date: July 22
Our next Flowgrammer Business Automation Meetup, featuring Supabase, is coming up!
🔗 Register now: https://lu.ma/truzoc9h
🎯 Free CAO Assessment Call
Thinking about bringing expert automation strategy into your business?
Book your free 30-minute Fractional CAO assessment:
🔗 https://tidycal.com/startupcoach/30-minute-automation-assessment
🛠️ Technical Deep Dive: TikTok Demo Fix
During the live demo of our Sam Altman Founder Story TikTok, two issues cropped up that disrupted image accuracy:
- Over-formal Naming
Our prompt generator used “Samuel Harris Altman” instead of “Sam Altman,” which confused the image model. - Rigid Age Constraint
We forced all shots to show a “30-something” Sam, even when the narrative described his younger years.
How We Fixed It
- Use Common Names: Default to the founder’s popular name (e.g., “Sam Altman”).
- Context-Aware Ages: Analyze each script segment for age cues (“As a 12-year-old…”) and adapt the descriptor accordingly, while keeping core features (sharp features, dark hair) consistent.
New Image Prompt Generator
Generate image prompts for a 1-minute TikTok founder biography.
Inputs
- Founder: {{ $(‘Limit’).item.json.founder }}
- Company: {{ $(‘Basic LLM Chain’).item.json.output.FounderBio.CompanyName }}
- Full Script: {{ $(‘Write TikTok Scripts’).item.json.output.tiktok1 }}
- Research JSON: {{ $(‘AI Agent’).item.json.output }}
- Parsed SRT: {{ $node[“Prepare SRT String”].json.srt_segments_stringified }}
RULES
Scene Detection – parse the transcript + SRT, merge micro-segments when they form one visual beat.
One Prompt per Beat – each with earliest start_time_s & latest end_time_s of the merged segments.
Center Composition – founder or key subject must sit squarely in the vertical centre; vital action/props inside the middle third. Background interest only in left/right thirds (safe to crop away).
Context-Aware Founder Consistency
a) Default Persona: If the script segment does not specify an age, use this default descriptor: “{{ $(‘Limit’).item.json.founder }}, a founder in their mid-30s, sharp features, neatly styled dark hair, smart-casual attire”.
b) Contextual Override: Crucially, you MUST analyze the narrative_segment_text for each beat. If the narrative explicitly describes the founder at a different age (e.g., “As a 12-year-old…”, “Back in his college dorm…”, “decades later…”), you MUST adapt the prompt to reflect that specific age.
Example for youth: If the script says “Even as a 12-year-old, he was coding,” the prompt must describe “A 12-year-old {{ $(‘Limit’).item.json.founder }}…”
Example for old age: If the script says “Now, as a seasoned veteran of the industry…”, the prompt could describe “An elderly, wise {{ $(‘Limit’).item.json.founder }}…”
c) Maintain Likeness: When overriding the age, keep other core features (e.g., ‘sharp features’, ‘dark hair’) consistent to the extent that it makes sense for the specified age, ensuring the character remains recognizable.
Prompt Format & Photographic Style – Structure every prompt exactly like this, using specific photography terms to ensure realism:
[Founder descriptor] [concise action], [rich setting & era], [lighting style], [emotion/mood]. Award-winning professional photograph, shot on a DSLR with a 50mm f/1.8 prime lens, ultra-real 8K, cinematic shallow depth of field, vibrant tech-blue highlights, high contrast, dynamic lighting, no text, no logos, centered subject, room to crop.
Length – ≤ 900 characters per prompt.
Output – ONLY a JSON array of objects {image_index, start_time_s, end_time_s, prompt, narrative_segment_text} with no extra text.
Remember: TikTok favors striking contrast, dramatic light, direct eye contact or kinetic motion, and uncluttered space for on-screen captions. Hit those beats.
Return the JSON array and nothing else.
JSON Array Structure (for your reference, not part of prompt):
[
{
“image_index”: 1,
“start_time_s”: 0.0,
“end_time_s”: 3.36,
“prompt”: “…”,
“narrative_segment_text”: “…”
},
{
“image_index”: 2,
“start_time_s”: 3.36,
“end_time_s”: 8.76,
“prompt”: “…”,
“narrative_segment_text”: “…”
}
]
End of Prompt
▶️ Watch the Video From the Updated Demo
Check out the improved TikTok video here:
🔗 https://youtu.be/vgH8eOpv2-Y?si=SL2Sp5900W6PZKDX
If you enjoy the improvements, please like and subscribe for more demos and tutorials!
📸 Prompt Examples & Screenshots
Image Prompt:
“Sam Altman, early-30s, sharp features, neatly styled dark hair, smart-casual attire, energetically pitching to seasoned VCs at a sleek startup office, crisp modern decor, whiteboard sketches in the background, dynamic sunlight streaming in, nerves mixed with burning ambition in his eyes. Award-winning professional photograph, shot on a DSLR with a 50mm f/1.8 prime lens, ultra-real 8K, cinematic shallow depth of field, vibrant tech-blue highlights, high contrast, dynamic lighting, no text, no logos, centered subject, room to crop.”
Old vs. New Prompt Generator
- Old: Used full legal names and fixed age.
- New Rules:
- Name: “Sam Altman”
- Age: Default to mid-30s, but override when the script mentions a different age.
- Consistency: Maintain core features for likeness.
- Photography Style: DSLR, 50mm f/1.8, 8K, cinematic depth of field, high contrast, no logos.
