Inspiration
OmniFlow started as a class project but quickly evolved into something real. While building OmniFlow Advisory, a consulting firm helping creators and small business owners automate operations with AI, I noticed a recurring issue: most tools were fragmented, inflexible, and built for devs, not for the people running the businesses.
That insight led to OmniCore: a modular, AI-powered backend system that gives users the ability to connect their tools, automate logic, and scale workflows — all in one place, without needing to code from scratch. Think of it as an AI-native productivity engine designed for lean teams that want to do more with less.
What it does
OmniCore is a smart backend productivity OS for automation-first businesses. It allows users to: • Automate workflows using GPT, Zapier, Notion, and other APIs • Deploy intelligent modules like “Client Intake Sorter,” “AI Proposal Builder,” or “Insight Reporter” • Store and recall business logic with persistent memory and contextual awareness • Connect frontends like Bolt, Framer, or even Notion to make the system feel native
It’s designed to function like an AI teammate — managing tasks, parsing inputs, and triggering flows, all with a conversational interface and human-like reasoning.
How I built it
• Frontend: Bolt for rapid prototyping + Framer/React for interactive UI
• Backend: Node.js + Supabase for user management, task logic, and storage
• AI Layer: OpenAI API + LangChain for chaining logic and generating intelligent responses
• Automation: Zapier for triggering outside actions (email, calendar, CRM tools)
• Database: Supabase (Postgres) for fast and scalable storage
• Cloud: Google Cloud Platform + Vercel for deployment
Built using a modular service architecture so that each feature can scale independently.
Challenges I ran into
Designing a flexible UI that doesn’t overwhelm users • Handling multiple tool integrations with varying APIs and rate limits • Implementing memory and context chaining in a way that feels natural • Staying focused on the core use case when there’s so much AI hype noise • Managing both Capstone deadlines and real-world use case demands from early test clients
Accomplishments that I'm proud of
• Turned a solo Capstone into a working prototype with real client interest
• Created a flexible architecture that can grow into a full SaaS platform
• Got early feedback from creators, consultants, and solopreneurs who said this could replace multiple tools they currently juggle
• Proved that AI agents don’t have to feel like chatbots — they can be deeply useful backend teammates
What I learned
• Human-centered AI is what drives adoption, not raw model power
• APIs are powerful, but without logic orchestration, they’re just fragments
• Persistence + modularity = scale. Building OmniCore this way makes it extensible without complexity
• The hardest part isn’t building. It’s positioning and explaining AI systems in a way that clicks
What's next for OmniFlow
• Launch a private beta for solopreneurs and agencies using the full OmniCore stack
• Build a visual logic builder (drag-and-drop flow creation with GPT agents)
• Evolve the consulting arm (OmniFlow Advisory) into a service+product hybrid
• Expand AI agent capabilities to include memory persistence, scheduling, and real-time monitoring
• Most importantly, find the right teammates to take this vision to the next level
Log in or sign up for Devpost to join the conversation.