Decision Analysis Service Launch
November 4, 2025
I'm tired of working alone I spent eighteen months building a tourism app designed to improve front-of-house conversations with guests. Building tools for human connection. Alone in my apartment. Talking to AI. The irony is not lost on me. That’s why I'm now launching something that forces me to talk to actual humans during the workday: Decision Analysis.
Six weeks ago, I started doing systematic research into AI coding development methodologies. I wanted to solidify what I’ve learned building QR Cards. What did I get right? What should I refactor? What libraries and frameworks should I consider? After researching how different prompting methodologies perform for different types of programming problems (as shown at Puget Sound Programming Python “PuPPy” Meetup and AI Tinkerers, see spawn-experiments: https://github.com/ivantohelpyou/spawn-experiments/), I realized that the real frontier wasn’t writing code from scratch but rather figuring out which libraries to use. For example, let’s say you’re adding fuzzy search to your app. If you’re not careful, your AI agent may select FuzzyWuzzy when you should really be using RapidFuzz. Or even worse, you may end up with FloozyWoozy with malware that steals your data and your wallet. That led to a new project (“spawn-solutions”) researching which libraries work best in which situations. First, I came up with a Dewey Decimal System-inspired taxonomy of algorithms. Then, I started filling out the library: sorting & searching, compression & encoding, geometric & spatial. Now, when I want to build something, I pick a library, any library, and then use the appropriate method from spawn-experiments to build a wrapper for it. And guess what? The experiments work! My building is about to replace its elevators. We’ll be down to one elevator for an entire year. I recognized this problem. About 28 years ago as an MBA student at Owen, I took a single half-semester Operations module which taught us how to model a queue: λ for arrival rate, μ for service rate. We did a case study in Excel and (other than every single time I’m waiting in line) I haven’t thought about it since. My spawn-solutions research into discrete event simulation (DES) libraries pointed me to salabim, which includes built-in animation, a lower learning curve, built-in statistics, and real-time mode for interactive demos. From there, I quickly figured out how to model traffic patterns for a simulated elevator replacement project. Please note that previously, I had never heard of DES or salabim, let alone built an animated simulation. According to my calculations, residents will experience 75% longer waits, with some waiting up to 30 minutes during peak usage periods. As a mitigation, voluntary “Express Stops” (every five floors) would make service much more predictable with faster service and much fairer wait times (<14 minutes max with 58% reduction in stdev). And as I get better estimates for traffic patterns – commute timing, deliveries, trash disposal – my model will improve. But it’s already good enough to spark a lively discussion. Think about what this means. If I can model an elevator in a day, anyone who really knows what they’re doing will be able to model a retail store or a convention center or an airport or even a city, much faster and with much less friction than ever before.
I realize that most people aren’t building elevator simulators in their spare time.
Businesses have more pressing problems these days. Either you have a stack of disconnected SaaS products with overlapping functionality, or you have an expensive suite with high lock-in and non-optional AI “upgrades.”
And so I started building a decision framework for purchase decisions, with systematic research into major categories: accounting, payment processing, financial planning, event management, CRM, ERP.
Using this framework, I want to help business owners make critical decisions:
Can you do it with the tools you have?
Maybe you don't need new software. Maybe you just need to use what you've got differently.
When you buy, what software/services are best for your business?
Stripe vs Paddle? QuickBooks vs Xero? Sentry vs Datadog?
How can you enhance the stack you have using AI coding?
This is the fun part.
Example: You have QuickBooks. You’re wondering about how different tariff scenarios will affect your business. Do you buy FP&A software that integrates with QuickBooks? Do you upgrade to more powerful accounting software? Or do you export from QuickBooks and build a custom scenario model. Which library? Which approach? That's what I research.
Now instead of sitting alone doing research, I want to talk to actual humans about their actual problems. And if my research can help you get started, it’s worth the conversation. Pull up a chair, I’ll pull up an agent, and we’ll see what happens.
Book a free call: app.ivantohelpyou.com/decision-analysis