Why “Sht in, Sht out” perfectly sums up the power — and limits — of generative AI.
AI is everywhere — promising radical efficiency, stirring existential debates, and flooding your feed with new tools every week. But hype is easy. Making it useful is harder.
At LDSK, we don’t ask if AI is here. We ask: how do we make it actually useful — for our teams and for our clients? What practical value can it deliver today, not just in theory?
One line that’s stuck with us — and gets truer by the day — comes from Josef Schneider, who captured it perfectly:
“AI won’t replace you, but somebody using AI will.”
It’s more than a soundbite — it’s a mindset. And for us, it’s core to how we build, work, and deliver.
This post pulls back the curtain on one real use case inside our marketing team: creating promotional screen content for our Samsung display at ISE — a high-pressure, short-turnaround project where generative AI helped us move fast, adapt quickly, and deliver under pressure.
From Concept to Content: AI on Deadline
When Samsung invited us to feature in the PIRS (Pre-Integrated Repeatable Solutions) section at ISE — the world’s leading pro AV and systems integration show — we jumped at the opportunity. Our intelligent media scheduling platform would be featured live, but we needed something more: engaging mock content that demonstrated how LDSK handles programmatic-style advertising in real-world retail environments.
The constraints were clear:
- No real client media (licensing restrictions)
- High-quality visuals to do justice to Samsung’s 4K displays
- A very tight deadline
So we asked ourselves: How far can AI take us — practically — in content creation under pressure?
Our AI Workflow: Tools and Takeaways
We built a rapid AI-powered pipeline, assigning each tool a distinct role:
1. Ideation – ChatGPT & Gemini
We used prompting to brainstorm mock product categories like “luxury wellness” and “smart home gadgets,” generating ideas and visual references on the fly.
Example prompt:
“Create a digital artwork with a luxurious theme. A tall, black cylindrical bottle with a silver cap, placed on a reflective surface that shimmers.”
Some results were striking. Others? Think flying teacups or extra limbs. But for early-stage creative input, it worked.
2. Image Refinement – Adobe Photoshop (Generative Fill)
Rather than re-prompting endlessly, we used generative fill in Photoshop to clean up weird artefacts and adjust layouts fast. It gave us far more control — and speed — over the final look.
3. Motion Design – Imagine New
We added life to static assets using Imagine New — animating subtle movements to catch attention onscreen. It wasn’t perfect with people or complex scenes, but for mock product ads? It delivered.
4. Final Assembly – Canva
Brand elements, labels, and layout polish came together in Canva — ready for display on the Samsung stand as mock campaigns running through LDSK’s dynamic scheduling system.
Our outcome…
Lessons Learned: “Sht In, Sht Out” Is Real
Generative AI helped us ideate and execute quickly — but precision? That still required human input. We learned (and re-learned) that:
- Good outputs require good inputs. Vague prompts = vague results.
- Domain knowledge matters. You need to know what “good” looks like to guide AI toward it.
- Human oversight is non-negotiable. From glitchy renders to brand tone — AI is your co-pilot, not your replacement.
Since then, we’ve begun experimenting with tools like Flora.ai to help us structure better prompts, maintain creative context, and streamline AI-assisted workflows.
It’s early days, but the trend is clear: The more intentional we are upfront, the better the output.
AI Across LDSK: Not Just for Marketers
While this ISE project came from our marketing team, LDSK’s use of AI spans far wider:
- Developers use AI to prototype, iterate, and debug faster.
- Product and marketing teams leverage Gemini and ChatGPT for research, writing, and automation.
- Meetings are supported by AI-driven note takers like Google’s, capturing action points without distraction.
- FlowSavvy helps manage time, tasks, and cross-team priorities more intelligently.
Everywhere we use it, AI removes friction and frees up focus — helping people spend more time on the work that matters.
Deterministic vs Generative: Knowing the Difference Matters
The tools we used for the Samsung stand were generative — brilliant for speed, exploration, and creative direction. But they also come with unpredictability.
By contrast, LDSK’s core platform is powered by deterministic AI. It makes precise, scalable media scheduling decisions — with clarity, logic, and no hallucinations.
Here’s how we think about it:
- Generative AI is your imaginative creative director — fast, expressive, sometimes chaotic.
- Deterministic AI is your calm, consistent engineer — logical, dependable, and made for scale.
Both are valuable. The trick is knowing when to use which.
Final Thought: Specific In, Strategic Out
If there’s one thing this project taught us — and keeps teaching us — it’s this:
AI is not magic. It’s a multiplier.
Put in vague prompts, half-baked ideas, or generic thinking? That’s what you’ll get back.
But when you bring clear inputs, domain expertise, and thoughtful guidance, the results can be transformational.
At LDSK, that’s our goal: to use AI not to replace people, but to help people do better work — more strategically, more creatively, and more efficiently.
Want to explore how LDSK’s AI can support your media planning and scheduling workflows?
Let’s talk. Whether you own, run, or manage media, we can help you move faster, target smarter, and plan with precision.