About Us
Hi, my name is Grady, and I'm the founder of xplo.
The Problem
Several years ago, as part of some non-profit work I was doing, I was working to modernize a project called OPGEE - at the time, it was a spreadsheet that estimates the greenhouse gas emissions that come from oil production itself, with the aim of helping to minimize the carbon impact of the extraction process (even oil producers can minimize their impact).
OPGEE was built by a bunch of really smart researchers who understood the chemsistry and mechanical processes of oil production, and had built a spreadsheet to help calculate their emissions. The third version (OPGEE v3) was a 23MB spreadsheet and took a long time to open, let alone actually run! Our goal was to make it faster, and have it run not just in Excel, but make it widely accessible and useful in data pipelines and software systems. The effort of porting OPGEE v3 to the OPGEE v4 python library took a team of researchers months of engineering time, and distracted them from their core work of understanding carbon modeling.
The idea
From that experience I got two really powerful lessons:
- Excel is how a lot of people like to do their modeling, logic, and collaboration
- Excel is VERY hard to use inside other processes like data pipelines or APIs
There had to be a better way - there were too many great ideas trapped in spreadsheets, and too much brainpower going to translating between Excel and other systems.
With that goal in mind, I started building xplo - a mechanical layer that translates an Excel spreadsheet into fast, reliable code.
Xplo's aim is to help empower the (large) segment of the world that understands and does their thinking in Excel - then helps connect their work into any other system.
Our Broad Vision
Xplo is built by the team at Pythia Software, a Colorado LLC. Pythia's overarching goal is to solve a generic class of problems called 'oracle problems'. An Oracle problem is one where you have access to what the "right" answer is, and are trying to build a systematic/faster way to get to the right answer.
LLMs make approaching oracle problems (even big ones, like the excel engine) much easier. Xplo has been our first project in this direction - we've built a robust, bulletproof, deterministic system, by leveraging large volumes of carefully orchestrated AI. Through xplo as a first example, we've built out reusable infrastructure that will allow us to tackle further generic oracle problems in the future - systems like data migrations and infrastructure replacement that should allow folks to replace and build around their older pieces of software.
We think LLMs are going to be transformative in software, not just in allowing people to build more of it, but letting us constantly upgrade and improve on the speed, efficiency, and compatibility of our systems, without human effort. Xplo is our first bet on that thesis.