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
Connecting 4 AI agents in Microsoft Teams to handle your entire retail operations? Yes, it's possible. 🤖💬
From Store SOPs to Corporate Admin, HR, and Ops—we built a multi-agent workspace powered by LLMs and a smart knowledge base.
Here is exactly how the architecture works under the hood:
📦 Company Docs (PDFs, Word, SOPs)
⬇️
🧩 Chunked & Embedded into a Vector Database
⬇️
💬 User asks a question right inside Teams
⬇️
🔍 Bot searches Vector DB for the exact context
⬇️
🧠 Claude/GPT synthesizes the precise answer
⬇️
✅ Verified answer appears instantly in the chat
No more digging through messy shared drives. Just instant, compliant answers for store managers and floor staff. 🚀
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
Mainframe is a conceptual AI studio — brand system + full multi-page website, built entirely in Stitch. Landing page with animated states, five inner pages (Labs, Studio, Services, Openings), and a complete DESIGN.md covering color, type, and layout. Monochromatic, editorial, cinematic. The kind of system that usually takes weeks — done in one session.
Started with brand tokens, used them as the foundation for every page. Iterated hover states and animations without leaving the design context. The generated DESIGN.md became the actual handoff doc.
Fastest I've gone from concept to a coherent multi-page system. Want: auto-propagating token updates across pages, finer motion controls, native multi-page navigation. Otherwise — best design-gen tool I've used since Figma got auto-layout.
Sahil, I recall seeing that same exact monitor-head concept from a large design studio last week (I forget the name).
NeuralOS The AI-Native Operating System
NeuralOS is a concept for a next-generation operating system built around autonomous AI agents rather than traditional applications. Instead of manually switching between tools and workflows, users simply provide a goal, and the system coordinates specialized agents to research, analyze, organize knowledge, and generate complete deliverables.
At the center of the experience is the Intelligence Core, a living system that orchestrates communication between agents, manages knowledge networks, and transforms user intent into actionable results.
The project explores a future where operating systems become intelligent collaborators rather than passive software environments.
NeuralOS is a conceptual AI-native operating system designed around autonomous intelligence rather than traditional applications. Instead of manually navigating between different tools, users provide a goal, and the system coordinates specialized AI agents to research information, analyze data, organize knowledge, and generate complete results. At the center of the experience is the Intelligence Core, which acts as the operating system's brain, orchestrating communication between agents and managing the flow of information across the platform.
How I Used Google Stitch
Google Stitch was used throughout the entire design process, from initial ideation to final refinement.
My workflow included:
Generating the initial operating system concept
Exploring multiple layout and interaction directions
Refining screens through iterative prompting
Building a consistent visual language across the product
Designing complex agent orchestration workflows
Creating a connected multi-screen user journey
Rapidly testing and evolving ideas without rebuilding screens from scratch
Stitch allowed me to move from a rough concept to a complete product vision significantly faster than a traditional design workflow.
Design Goals
The challenge theme was "Build interfaces that feel alive."
To support this theme, NeuralOS focuses on:
Living intelligence systems
Dynamic agent collaboration
Spatial knowledge visualization
Real-time orchestration workflows
Motion-driven feedback
AI-native interaction patterns
Rather than creating another dashboard, I wanted the interface to feel like an intelligent ecosystem that continuously thinks, learns, and evolves.
## Feedback on Google Stitch
My experience with Stitch was extremely positive. The platform made it easy to rapidly explore ideas, iterate on layouts, and build a cohesive product experience across multiple screens.
One area I would improve in future projects is the use of generated icons and decorative assets. While they are useful during ideation, I found that replacing them with established icon libraries and design systems creates stronger visual consistency and makes the final product feel more realistic and production-ready.
The strongest results came from combining Stitch's AI-assisted generation with thoughtful human refinement. This balance allowed me to move quickly while still maintaining creative control over the final experience.
GitHub Repository:
https://github.com/MHK-123/NeuralOS
Google Stitch Project:
https://stitch.withgoogle.com/projects/1582384152650609720
Built a RAG job-hunting agent for Indonesian developers applying to remote USD jobs.
28 days of real usage. Here's what I learned.
The problem: applying to 50+ remote jobs manually is exhausting. Wrong roles, no resume tailoring, hours wasted per application.
The workflow now:
One message → agent retrieves relevant jobs → ranks them → generates tailored resume + cover letter → user approves or skips.
Stack (still running $0 free tier):
· Embeddings: Cloudflare Workers AI qwen3-embedding-0.6b
· Vector search: Cloudflare Vectorize, cosine similarity
· Database: Cloudflare D1 / SQLite, one batched query
· Semantic cache: Cloudflare KV, cosine ≥0.92 threshold
· LLM: JSON-mode slot extraction + reply
Only variable cost: ~$0.002/turn LLM usage
28-day numbers:
· 235 sessions
· 329 return visits
· 743 agent turns
· 357 job cards surfaced → 46 approved applications
~20 approvals per 100 sessions (target was 10–15)
Total LLM cost: ~$1–4
Example matches from one natural-language query:
Linear — 96/100 · DatAds — 95/100 · Spotify — 100/100
Two decisions I'd make again:
· The scorer is deterministic — pure function, 7 weighted components, 0–100. No model drift. Users see exactly why a job scored highly. I didn't trust LLM ranking for something this consequential.
· There's a regex fallback behind every LLM extraction. If the model fails or times out, the conversation continues. Users never hit a dead end.
What's next:
· Tracking downstream outcomes — recruiter replies, interviews Detecting expired/duplicate posts more aggressively Calibrating scoring weights from real approval behavior
· The pipeline runs cheaply and generates real approvals. The next challenge is proving it improves actual job outcomes — not just application volume.
This is the kind of system I enjoy building: AI-native product on constrained infra, with real users and measurable north-star metrics.
If you're working on something similar — hiring pipelines, talent matching, AI agents for job seekers — I'd enjoy comparing notes. And if you're an Indonesian dev hunting remote USD roles, the product is live at lokerdollar.com.
What would you build differently? 👇
#RAG #CloudflareWorkers #BuildInPublic #AIEngineering #RemoteJobs #IndonesianDevelopers
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?