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
Stop building AI features that require App Store updates.
I was asked to build 2 internal AI chatbots. I knew the org would eventually want 20. Instead of hard-coding them, I engineered a Server-Driven UI.
The Result:
Zero Deployments: Scaled from 2 to 12+ bots without a single app update.
90% Faster: "Time-to-live" dropped from days to minutes.
Secure: Managed access via a custom Airtable-to-n8n vetting system.
I turned a coding bottleneck into a dynamic plug-and-play hub. Build for the scale you expect, not just the one you have.
#Flutter #AI #Automation #SoftwareArchitecture
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
Related posts
87% of marketers use AI in at least one workflow now.
Only 6% have fully implemented it (Supermetrics 2026).
The gap between those numbers is basically the entire job description for what I do.
Going to break down what an AI automation engineer is, what we ship, and how to tell the bad ones from the good ones.
An AI automation engineer connects AI to the workflows your business already runs.
We don't train models. We don't write papers. We don't do research.
We pick the right model for the job, wire it into the systems your team uses every day, and ship the build.
17 months ago, we launched an AI chatbot for a small business client.
Here's an honest update.
Two months in, the numbers were modest.
80 conversations. 708 messages. 14% conversation-to-lead rate.
Not bad. But most visitors still preferred the contact form.
So we left it running.
We watched how people used it and made improvements. Nothing major.
17 months later:
1,350 conversations. 179 leads. 13.2% conversion rate.
At 16x the volume, the quality held almost exactly.
But there was a real shift around July 2025 that’s worth pointing out.
1. Average chats per month doubled, from 40 to 79.
A clear sign that where conversations are taking place is changing. People are no longer browsing sites in the same way.
They want their questions answered fast, and if you can’t deliver that, you miss out.
2. The chatbot became the #1 lead source on the site
From July 2025 onwards, the chatbot has consistently driven more leads than the forms, and that's held through 2026.
3. The average conversation runs 9 messages deep.
These are not quick FAQ lookups.
Some conversations span materials, pricing, customisation, planning permission and delivery.
These are real pre-sales conversations handled without a team member in the room.
PS: Only 3 thumbs down in 17 months. We'll take that.
Good work. Yeah as the industry shifts customer behavior does so as well
Atelier 360 — A Reusable Fashion Launch System
Full Project Description
Atelier 360 is a reusable creative workflow built in Melius that turns a single fashion product brief into a complete omnichannel launch campaign.
For this challenge, I used a premium black oversized hoodie as the starting product. From that one brief, the workflow generates brand direction, product-only e-commerce assets, 360-degree product views, on-model editorial visuals, social creatives, paid ads, out-of-home campaign concepts, website and email banners, and cinematic video direction.
The main idea behind Atelier 360 is product consistency. Generative fashion workflows often create beautiful isolated images, but the product can easily drift between outputs. To solve this, I built the workflow around a Canonical Product Identity and Global Quality Standards. These act as a source of truth for the garment’s silhouette, material, proportions, construction, and visual style.
The system is designed to reduce common AI issues such as inconsistent product details, exaggerated garment proportions, duplicate hoods, faceless models, ghost-mannequin artifacts, unreadable text, and off-brand visual drift. Instead of treating each image as a separate generation, Atelier 360 works like a connected production pipeline.
Key Functions & Capabilities
Strategic Brand Foundation
The workflow starts by generating a clear brand strategy, audience direction, visual tone, and campaign positioning based on the initial product brief.
Canonical Product Anchor
A central product identity node defines the source of truth for the garment, including silhouette, material, hood structure, pocket placement, cuffs, hem, and overall construction.
Product-Only E-Commerce Assets
The system creates clean product visuals for online stores, including hero product shots, 360-degree front/side/back views, material close-ups, and PDP-ready assets.
On-Model & Editorial Assets
The workflow separates model-based visuals from product-only assets to create editorial campaign images with visible faces, realistic anatomy, and premium fashion styling.
Omnichannel Marketing Assets
Atelier 360 generates campaign concepts for social media, paid ads, website banners, email banners, billboards, and street posters.
Video & Motion Direction
The system includes a video pipeline covering script, storyboard, keyframes, 360-degree product turntable concepts, and short-form fashion promo video direction.
Process Steps & Execution Strategy
I started by building the strategy layer and defining the visual direction for a premium minimalist streetwear brand. Then I created a Canonical Product Identity to keep the hoodie consistent across all future outputs.
From there, I organized the canvas into five functional zones:
Strategy & System Rules
Product-Only Assets
On-Model & Editorial Assets
Campaign & Marketing Assets
Video & Motion Direction
During the process, I learned that a large nested workflow is easier to control when it is executed group by group rather than all at once. This allowed me to review each stage, refine weak outputs, and keep the final campaign board cleaner and more consistent.
Feedback on Melius
Melius helped turn the creative process from isolated prompting into a visual production system. I liked being able to structure the workflow spatially, connect strategy to visual outputs, and use the agent to expand one product idea into many connected campaign assets.
The canvas format made it easier to think like a creative director: define the product, build the visual world, generate assets, review quality, and curate the strongest outputs. For me, the biggest value was not just generating visuals, but building a reusable workflow that can be adapted for future fashion launches.
this is so cool!
Trending
Claude
Claude has entered the design space. How are you using Claude Design?
Contra University
Learn from expert creatives how to earn more using next-gen AI tools.
creativeaiflow
Creative AI workflows are evolving. What tools do you use, and what are their strengths and weaknesses?
portfolioreview
The best portfolios tell a story, not just show a grid. Share yours for feedback.
freelancerlife
Freelancer life is wins, pivots, and everything in between. What’s yours right now?