Converting Excel Spreadsheets to Ruby on Rails Applications
Published July 11, 2026
Build in Public - GTM
In November 2025, my friend Darren that I met in San Francisco had a problem. His family business RSB Contracts, had this massive operating spreadsheet for doing all of their tendering of large commercial steel construction projects, and his brother, Richard, was looking for a way to simplify this process.
Since Darren is a true Excel wizard, and a management consultant, his brother asked for some help. Rich had been using ChatGPT to help him create Macros and VBA code to automate some of the more tedious setup tasks involved with the spreadsheet, such as getting BOQ data in the right format, so that the company's quantity surveyor, Demi, could do the tender.
They also implemented a bunch of lookup tables, automated templating, etc. But, they still had issues and the automations would often stop working.
Rich was at a party, and one of his friends who worked at a large company was talking about building a custom business software system for their corporation to solve some of these same problems. Rich got the contact information of the contractors who built it, and reached out. They spent months talking about the project with this business partners, and eventually got a quote from these contractors, which was over $20K USD for a slimmed down version of what he was wanting.
Eventually, Darren reached out to me, asking if I was still working on using LLMs to generate application code, and I told him I was. By this point, Darren had gotten really familiar with Rich's Excel sheets, and understood a lot of the business logic himself, and we started hanging out and building stuff for fun.
The first thing we built was a BOQ parser, where you could upload a CSV with BOQ materials, and it would send it to GPT Codex, and respond with structured output in the exact parsed format that made it easy to copy and paste into the Excel Tender template.
From there, we realized we could simply just recreate the Tender line items, and have Claude Haiku put their same Excel formula logic exactly in Ruby code by using Leo, an AI coding agent I'd developed the prior summer that generated Ruby on Rails code.
From there, we started building in better tools for Leo to read Excel sheets, understand business logic, and Darren began using Leo to build out his brother's custom software system. This Excel sheet had over 15 years of hard won lessons encoded in it's formulas and cells. Rich described it as "the brains" of the business, and by putting it into a proper web application and incorporating AI features such as the BOQ parser, this new custom business software was allowing their team to move much faster.
This is how we came to realize, that Excel is a powerful starting point for building custom business applications, and it's why we've built Leo to become a wizard at understanding Excel and building rapidly with business owners.
If you're curious to learn more, upload your Excel sheet to see how Leo reacts to it by going here: https://llamapress.ai/excel-to-app
Since Darren is a true Excel wizard, and a management consultant, his brother asked for some help. Rich had been using ChatGPT to help him create Macros and VBA code to automate some of the more tedious setup tasks involved with the spreadsheet, such as getting BOQ data in the right format, so that the company's quantity surveyor, Demi, could do the tender.
They also implemented a bunch of lookup tables, automated templating, etc. But, they still had issues and the automations would often stop working.
Rich was at a party, and one of his friends who worked at a large company was talking about building a custom business software system for their corporation to solve some of these same problems. Rich got the contact information of the contractors who built it, and reached out. They spent months talking about the project with this business partners, and eventually got a quote from these contractors, which was over $20K USD for a slimmed down version of what he was wanting.
Eventually, Darren reached out to me, asking if I was still working on using LLMs to generate application code, and I told him I was. By this point, Darren had gotten really familiar with Rich's Excel sheets, and understood a lot of the business logic himself, and we started hanging out and building stuff for fun.
The first thing we built was a BOQ parser, where you could upload a CSV with BOQ materials, and it would send it to GPT Codex, and respond with structured output in the exact parsed format that made it easy to copy and paste into the Excel Tender template.
From there, we realized we could simply just recreate the Tender line items, and have Claude Haiku put their same Excel formula logic exactly in Ruby code by using Leo, an AI coding agent I'd developed the prior summer that generated Ruby on Rails code.
From there, we started building in better tools for Leo to read Excel sheets, understand business logic, and Darren began using Leo to build out his brother's custom software system. This Excel sheet had over 15 years of hard won lessons encoded in it's formulas and cells. Rich described it as "the brains" of the business, and by putting it into a proper web application and incorporating AI features such as the BOQ parser, this new custom business software was allowing their team to move much faster.
This is how we came to realize, that Excel is a powerful starting point for building custom business applications, and it's why we've built Leo to become a wizard at understanding Excel and building rapidly with business owners.
If you're curious to learn more, upload your Excel sheet to see how Leo reacts to it by going here: https://llamapress.ai/excel-to-app