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
Most teams adopting AI are doing it wrong.
They install Copilot or ChatGPT Enterprise, call it "AI transformation," and wonder why productivity hasn't changed 6 months later.
Here's what actually works — the 3-layer AI stack I implement for engineering teams😍
Layer 1: Context (Claude Code + CLAUDE.md)
Your AI needs to understand YOUR codebase, not just code in general. CLAUDE.md files give it your architecture, conventions, and team patterns. This is the difference between an AI that writes generic React and one that writes YOUR React.
Layer 2: Orchestration (Agentic pipelines)
Single-turn AI is a calculator. Agentic AI is an employee. I build multi-step workflows that can research, draft, review, and ship — with human checkpoints at the right moments.
Layer 3: Knowledge (RAG + Karpathy Wiki Pattern)
Your institutional knowledge lives in Slack, Notion, and people's heads. I build structured knowledge bases that make that context queryable by your AI systems.
The teams I work with see 20-40% velocity improvements in the first 30 days. Not because AI is magic — but because context + orchestration + knowledge makes AI actually useful.
What layer is your team missing?
claude.md
Claude Code overview - Claude Code Docs
Claude Code is an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools. Available in your terminal, IDE, desktop app, and browser.
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
How are you actually structuring your AI workflow right now - are your tools connected, or are you still using them separately?
I’ve landed on a pretty simple system using just Claude + Flora.
Claude is where I develop the thinking - my existing brand direction, concepts, and refining prompts until they actually reflect the creative intent behind the project.
Flora is where that intent becomes visual. It’s a node-based canvas, so I can generate imagery, iterate variations, drop in mockups, and build full product scenes in one connected system.
I don’t see AI replacing design thinking. I see it speeding up execution while I stay fully in control of direction, taste, and decisions.
Curious to hear from you guys 🤔
📢 What does your AI workflow look like right now?
📢 What tools are you actually using day to day?
📢 And is there anything you think I should check out?
Would love to hear how others are building their systems.
this is wonderful, i love your lightning!
What if your AI workflow is the reason your brand still looks generic?
Q2 forced me to reset how I build everything creatively and honestly… I simplified it way down.
Right now my process is just two tools, but waaay more intention behind them.
I’m using Claude to actually think with me. I feed it strategy, references, and the full brand world, and it translates that into really detailed prompts. When you pair that with Nano Banana 2, you finally start seeing visuals that actually feel like your brand.
I stopped expecting AI to figure it out and started directing it like a creative partner.
That’s also why I moved away from MidJourney. The second you need anything brand-specific like real text, packaging, or logos, it starts to fall apart. It’s great for vibes, not for building real brand systems.
This workflow forces more clarity upfront, and the output is 10x more intentional.
Curious how everyone else is using AI right now… are you directing it, or letting it lead?
*Concept only. Not an official collaboration.
good talk
I’ve been experimenting with how AI agents can collaborate instead of working in isolation.
For this project, I set up two agents on a shared canvas using @paperdesign:
→ One agent focuses on UI/UX design
→ The other turns that design into a working website
The interesting part isn’t just the output — it’s the interaction.
The design agent creates the system and layout for both desktop and mobile, and the build agent reads directly from the canvas and keeps the live site in sync.
There’s no traditional handoff. Everything evolves together in real time.
This felt less like using tools, and more like watching a system collaborate.
Curious to explore where this goes next.
Trending
Runway
AI video generation is exploding. What are you dreaming up in Runway?
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?