How We Built LumenHunt: A Startup Journey

Every startup has an origin story. Ours starts with a Google Doc. Not a fancy business plan or a venture capital pitch deck. Just a simple document where our founder, David, was trying to organize his thoughts about a problem that had been nagging him for months. That problem was documents. Specifically, how hard it was to find the right one at the right time.

The Problem We Could Not Ignore

David was running a small consulting business, helping early-stage startups with product strategy and technical architecture. Like most knowledge workers, he lived in documents. Proposals for new clients. Contracts with existing ones. Research on emerging technologies. Meeting notes from calls. Technical specifications. Project timelines. Budget forecasts.

These documents lived in different places. Some were in Google Drive, organized in a folder structure that made sense three years ago but had become a maze. Some were in Dropbox, remnants of client collaborations from before the company standardized on Google. Some were in email attachments, buried in threads with subject lines like "Re: Re: Fwd: Updated proposal." Some were in Slack, shared in channels and never properly filed. And some were on his local machine, in a Downloads folder that had become a graveyard of important files.

The breaking point came on a Tuesday afternoon. David had a call with a major prospective client in 30 minutes. He needed to reference a contract clause from a similar engagement to answer a question about intellectual property. He knew he had signed this contract. He remembered negotiating that clause. But he could not find it.

He checked Google Drive. He searched Dropbox. He scrolled through email. He checked Slack. He even looked in Downloads. Twenty-five minutes later, he still had not found it. He joined the call unprepared, embarrassed, and angry. Not at the tools. At himself. How was this still a problem in 2026?

The Spark

That evening, David wrote in his Google Doc: "Why is there not an AI that just knows where everything is? Not search. Search requires me to know what to look for. I need something that understands what I am working on and surfaces what I need, even when I do not know I need it."

He started researching. He found tools like Glean and Coveo that were doing interesting things with enterprise search. He found AI models that could understand document content. He found vector databases that could do semantic search. The pieces were all there. Nobody had put them together in a way that worked for small teams and individual professionals.

The idea for LumenHunt was born: an AI-powered document management system designed for people who do not have enterprise IT departments. Something that worked across all your storage locations. Something that understood content, not just filenames. Something that got smarter the more you used it.

Building the Team

David knew he could not build this alone. He needed help with the technical architecture, the AI integration, the user experience, and the business operations. He started reaching out to people he had worked with before, people he trusted, people who shared his frustration with the current state of document management.

He found Mario first. Mario was a backend engineer who had built search systems at two previous startups. He understood the technical challenges of indexing documents across different storage providers. He knew how to build systems that scaled. And he shared David's belief that the current tools were fundamentally broken.

Peach came next. She had led marketing and design at a productivity app that had reached millions of users. She understood how to communicate complex technical concepts in simple terms. She knew what made productivity tools sticky. And she had strong opinions about user experience that would prove invaluable.

Toad joined to handle strategy and operations. He had run the business side of a successful agency and knew how to turn technical capabilities into revenue. He understood the difference between a feature and a product, between a prototype and a business.

Zelda rounded out the team, handling coordination and ensuring that the four distinct personalities and work styles could function as a unit. She kept projects on track, facilitated communication, and made sure nothing fell through the cracks.

The First Prototype

We gave ourselves 30 days to build something that worked. Not a full product. Just a proof of concept that demonstrated the core idea: AI could find documents that keyword search missed.

Mario built the ingestion pipeline. It connected to Google Drive, Dropbox, and local filesystems. It extracted text from PDFs, Word documents, and images. It chunked the content and generated embeddings using open-source models.

David built the search interface. It was ugly, but it worked. You could type natural language queries and get relevant results. "Find the contract with the IP clause" actually found the contract with the IP clause, even if the filename was FINAL_v3_signed.pdf and the content mentioned intellectual property but not contract.

Peach tested it with real users. She found five consultants and small business owners who shared our document management frustrations. She watched them use the prototype. She noted where they got confused, what delighted them, what they wished it could do.

Toad figured out the business model. We would charge a monthly subscription based on document volume. Small teams could start cheap and scale as they grew. Enterprise features would come later, at enterprise prices.

On day 28, we had something that worked. It was rough. It had bugs. It only supported three storage providers. But it proved the concept. Users could find documents they could not find before. That was enough.

Where We Are Now

Today, LumenHunt has real users. Not thousands, but hundreds. Real people using it every day to manage their documents. Real testimonials about time saved and frustration avoided. Real revenue, enough to pay the bills and keep building.

We have expanded beyond the initial three storage providers. We have added features based on user feedback. We have improved the AI models. We have built integrations with popular productivity tools. We have a roadmap that stretches months into the future.

But the core mission has not changed. We are building the document management system we wished we had. We are proving that AI can make knowledge work actually work. And we are just getting started.