The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
RAG or fine-tuning? The question I get asked most and the answer is almost never "fine-tune."
Quick way to think about it:
→ Your data changes often, or you need the model citing real, current info? Go RAG. Faster to build, cheaper to update, and you're not retraining every time something changes.
→ You need the model to consistently think or respond in a very specific style/format that plain prompting won't hold onto? That's when fine-tuning earns its cost.
Most business use cases internal knowledge bots, support agents, research assistants are RAG problems wearing a fine-tuning costume. People reach for the expensive option because it sounds more "AI," not because it's the right one.
Building something and not sure which one fits? Happy to help you think it through.
#AI #MachineLearning #RAG #SoftwareDevelopment #Contra
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Trending
retrofuturism
Share your favorite designs from the past exploring the distant future. Or your own take on the aesthetic 🚀
halftone
Dots. Texture. Grain. Let's see your designs using the halftone technique.
Midjourney
Creatives are showing off what they're making with Midjourney.
colorgrade
Video editors sound off: How do you enhance the mood of your video in color grading?
Related posts
What does AI get wrong when you hand it a real brief?
I gave Kittl's agentic AI a brief we'd already shipped at Formeta, retail packaging for Ella by Ellaya Health, an at-home device for a common pelvic health condition, sold over the counter in Indian pharmacies. Plain language, no steering.
Pass one: four directions, three of them straight to the wellness reflex, lotus marks and sunrise gradients. All four were front faces attached to no box.
Pass two, with panel requirements and regulatory copy, came back with a real dieline. Glue flaps, directions for use, batch and expiry fields. Structurally a genuine leap.
It was also the wrong product. Invented a manufacturing licence number, added a CE mark to something sold only in India, moved the client to a different city.
On direction, Kittl is legitimately useful and I'd use it again. But healthcare packaging is a compliance problem wearing a design problem's clothes, and the model has no way to know it doesn't know that.
Vague brief gets you pretty. Precise brief gets you PLAUSIBLE. Plausible is the more dangerous one, because a founder without a designer might send it to print.
Last slide is the box we built.
Where do you let AI take the first swing, and where do you refuse?
It was super fun to experiment using Kittl with @Ansh Jamdagni 🫶
How often are you integrating Claude into your daily client work?
I’ve been pushing my AI usage beyond just basic ideation to see how it handles the heavy lifting in a real-world build. I recently completed a full project where Claude was my primary copilot from the initial architecture right through to execution.
You can check out the final deliverable here: AXIS Website
I’d love to hear how others are utilizing it. Are you using it for structural logic, content mapping, or something else entirely? Drop your workflows below. 👇
great work using claude!!
Same thing in engineering. Before I write a line of Core ML pipeline code I sketch the data flow on paper — where the tensor lives, where it gets copied, what actually runs on the ANE. The messy version catches the expensive mistakes while they're still free to fix. The keyboard is where you commit; paper is where you're still allowed to be wrong.
Is sketching part of your creative workflow or we've forgotten what it feels to be human?
Depending on the projec's scope, I normally like to explore my ideas on paper before I come up with the best solution.
And no, they're not perfect.
They are rough and messy, my goal is so...
