ORCFLO Stealth: The Beginning of our Journey

ORCFLO began as a passion project in 2025, consuming our nights and weekends...becoming a real company in 2026. All this happened in an unplanned 'stealth mode'. This series explains the back-story of our journey.

If you've spent any time with ChatGPT, Claude, or Gemini, you know the feeling. You type a question, hit enter, and watch something remarkable happen. The AI writes a compelling email in seconds. It summarizes a dense report in a paragraph. It explains a complex financial concept in terms your board would actually understand. It debugs code, drafts contracts, brainstorms marketing copy. These tools can do things that would have seemed like science fiction just a few years ago.

We Love AI - It's Just Too Hard to Use

We're believers. We use these tools every day, and they've genuinely changed how we work.

But we're also frustrated.

We're executives, strategists, analysts - people who've spent careers making decisions, managing teams, and navigating complex business problems. We're tech guys and know what these AI models can do - we use them all day every day.

We have evolved to super-users of LLMs and needed a way to string together a series of steps in a workflow. Without having to sit and do it all ourselves...step-by-step, over-and-over. What we needed was a way to orchestrate agents so they can handle a complex set of tasks while we work on other things. The problem? The simple chat interface.

Challenges With Chat

The chat interface is fantastic for one-off queries. Need to understand a competitor's latest earnings call? Ask Claude. Want to draft a response to a tricky client email? ChatGPT handles it beautifully. But when we tried to do something more ambitious - say, analyze a dozen potential investment targets using the same methodology, or build a repeatable process for evaluating job candidates - the chat model started to feel limiting.

We couldn't see what was happening behind the curtain. We had to do everything manually, step-by-step, waiting and watching. We couldn't easily compare how different models handled the same question, or understand whether we were overpaying for capability we didn't need. And when we wanted to refine our approach, to iterate on a process until it worked just right, we found ourselves copying prompts into Word documents and losing track of which version was which.

We Need Microsoft Excel For AI

When we have to work with a lot of numbers in columns and rows, spreadsheets are the answer. Excel has been our workbench for decades - we can model scenarios, tweak assumptions, and run analyses all day long. But for AI? All we had was a chat window. Type, wait, read, repeat.

We started looking for something better. A tool that would let us lay out a series of steps, assign AI agents to specific tasks, see what each step produced, and refine our approach over time. Something visual and intuitive. Something that didn't require us to write Python or tussle with APIs.

Nobody Has What We Need - Let's Build It Ourselves!

We couldn't find it.

Everything we encountered was built for developers. We're tech guys and CAN code, but we don't want to have to sit down and code every time we have some new problem to solve. We wanted something out-of-the-box that is easy and fast.

So we made a decision that, looking back, was either bold or naive. Probably both.

If the tool we wanted didn't exist, we'd build it ourselves. Our nights and weekends passion project was born.

NEXT UP:
Step 2: The Chat Model is Too Constraining