Inspiration - I was scrolling through X, formerly Twitter, when a post by a Chinese builder caught my attention. He described how he approached businesses with outdated websites by showing them a preview of how a modern website could help attract more customers. He built that workflow using Claude Code.
The idea stayed with me. The next day, I came across this hackathon and realized I could build something more complete, reliable, and functional around that concept. What started as a simple website preview and outreach idea quickly became something much bigger: a platform that helps businesses understand which prospects to approach, why they matter, what problems they have, and how to reach them with evidence-backed outreach.
That became OrbiVector.
What it does - OrbiVector started as a tool to find businesses with outdated websites, generate a modern website preview, and prepare an outreach email. But after working on it for several days, I discovered a much larger GTM problem.
Most businesses do not just need more leads. They need to know:
Why should we approach this client instead of another one? What problem does this business actually have? Why did a prospect reject us after the first meeting? Which leads are worth following up with? What should we say to them? Which channel should we use: WhatsApp, Gmail, LinkedIn, or something else?
OrbiVector solves this by acting as an AI agent swarm for local GTM intelligence.
It discovers businesses, verifies contacts, resolves websites, checks Google Business and Facebook signals, generates intelligence reports, ranks opportunities, and prepares personalized outreach across channels such as WhatsApp, Gmail, and LinkedIn.
For each business, OrbiVector can generate multiple intelligence layers, including:
Local SEO Intelligence Business Intelligence Competitor Intelligence Contact Intelligence Website Audit Intelligence Marketing Intelligence Company Intelligence Recipient Intelligence Outreach Intelligence Revenue Opportunity Intelligence Digital Maturity Intelligence Keyword Gap Intelligence Backlink Intelligence, planned through API integration Landing Page Creation Intelligence Landing Page Preview and Generation Intelligence
These intelligence layers are powered by coordinated AI agents. Each agent performs a specific role, passes its output to the next agent, and contributes to a complete GTM workflow. The result is not just a lead list. It is a verified prospect, a business report, an outreach strategy, and a ready-to-use action plan.
OrbiVector is built around the idea of AI agents working like digital employees.
How I built it - Building OrbiVector was one of the most enjoyable parts of the journey. I built a large part of it through vibe coding, experimenting quickly, testing workflows, and improving the system step by step.
Google Gemini is used as the primary AI model for intelligence generation. The platform is deployed on Vercel, with Supabase as the backend database and authentication layer. Google Cloud Console is used for Google-related authentication, verification, and integrations.
Over time, I integrated multiple intelligence and enrichment flows, including lead discovery, contact extraction, website resolution, report generation, Gmail draft creation, outreach planning, and dashboard analytics.
Challenges I ran into - The biggest challenge was not building one AI agent. It was making many agents work together reliably.
In theory, an agent swarm sounds simple: one agent completes a task, passes information to another, and the next agent continues the workflow. In practice, it was much harder.
At the beginning, out of more than 15 agents, only a few would run correctly. Some agents waited for manual commands. Some produced incomplete information. Some workflows broke when one agent did not pass the right data to the next one. I had to spend more than two weeks fixing coordination, data flow, task execution, and reliability.
The challenge was making the system behave less like disconnected prompts and more like a real team of AI workers.
That process made OrbiVector much stronger than the original idea.
Accomplishments that I am proud of - OrbiVector itself is my biggest accomplishment.
I started with a simple idea: find outdated websites and prepare outreach. It has now evolved into an AI agent swarm platform that can discover leads, verify contacts, generate reports, create outreach strategies, and prepare Gmail drafts.
As a single founder, I built a system where more than 15 AI agents work together like digital employees. Each agent has a role, and together they form a complete GTM intelligence workflow.
I am also proud that OrbiVector has already been tested across real internal workflows, including large lead generation runs, verified contact extraction, business intelligence reports, and outreach draft generation.
What I learned - I learned that building with AI is not just about asking an AI model to generate an answer. The real challenge is designing the system around the AI.
I learned how to benchmark AI agents, push them beyond simple prompt responses, coordinate multiple agents, validate outputs, and turn them into something closer to a digital workforce.
I also learned that GTM is not just about finding leads. It is about understanding which leads matter, what pain points they have, how strong their digital presence is, who their competitors are, and what message should be sent next.
That is the real value OrbiVector is trying to unlock.
What's next for OrbiVector - The next step is to make OrbiVector even more powerful and enterprise-ready.
I plan to add backlink intelligence through API integrations, improve competitive analysis, expand verified contact enrichment, strengthen reporting, and make the platform more useful for enterprises, SMEs, MSMEs, agencies, and local growth teams.
The goal is to make OrbiVector an affordable and reliable GTM intelligence platform that helps businesses move from market discovery to verified contacts, reports, ranking, and outreach.
Winning this hackathon would help validate the vision, secure support for future growth, and bring OrbiVector closer to becoming a serious AI agent swarm company.
Built With
- json
- node.js
- python
- supabase
- tsx
- typescript
Log in or sign up for Devpost to join the conversation.