The Script

In the first post, Building in Public I talked about "The Script" and "The Program." If "The Program" is the machinery -- the tech stack, the code, the automation -- then "The Script" is the narrative. It is the reason the machinery exists.

And right now, the page is blank.

The Engineer's Trap

When starting a company, I have to avoid a specific pitfall: The Engineer’s Urge.

The Urge tells me to build something complete, perfect, and scalable. It wants me to set up Kubernetes clusters, configure perfect CI/CD pipelines, and optimize for millions of users before I have a single one. A new product doesn't need a software engineering mentality, it needs a hacker mentality: get code out the door by any means necessary. Get ANYTHING onto a page, and edit later.

In the previous post The Program I mentioned one-way and two-way door decision making. In the early stages, the vast majority of decisions are going to be two-way doors. I need velocity. The goal is to get anything in front of a user to see if they care.

My mantra: Do it faster. Do it smarter. Do it better.

Finding The Wedge

Building without a problem is a recipe for wasted time.

The business side of things is not entirely foreign to me, but I am definitely an outsider here. Drawing on my past privileged experiences of midnight calls with IHS Markit's CEO, months of product positioning for Google product demos, and years of owning a billing system for an AWS product, I know that there are some basic things I need to figure out:

The Problem: A specific pain point.
The Industry: A market large enough to matter.
The ICP: An Ideal Customer Profile (who is actually writing the check?).
The Economics: Does the math work?

Case Study: Solar

In 2020, I started building something close to my heart: a tool called flexfarmd for enabling people to build renewable energy locally, track the deployment from start to finish, and optimize their generation. While working through the early parts of the user experience, I quickly ran into a wall: permitting.

Every city has their own permitting requirements, each permit needs dedicated support and work from experts, and someone has to shepherd everything through to approval. The hard part wasn't the software or the hardware, it was the bureaucracy. If software could shepherd that process, it would unlock the market.

I let the project sit while I focused on my work at AWS. In that time, competitors sprang up, got funded, and validated the market. They proved I was right. I still believe the space is a fruitful problem area, but the landscape has changed. I need a new wedge.

This is where I get to prompting.

Renaissance Developer

It's 2025.. We are writing this script in the age of the LLM. Coming from a creative background, I love to hate on AI hype, but I cannot deny the reality: AI is a force multiplier. I can audit an industry, generate user personas, and stress-test economic models in an afternoon rather than a month. I can generate quick UI mockups. I can confirm best practices for business, for code, and for content. I can feel it making me dumber, but the faster feedback loops are undeniable. Disclaimer: This does not mean that I do not need business advisors or UI designers, it means I can sanity test things and iterate through ideas before engaging experts. It's not a replacement, it's a prototyping tool.

How do we write the script? Audit industries, identify problems, define customers, test solutions. And ask questions.

I have a starting place (bureaucracy in construction), now I want to extend the themes (looking for a wedge in a validated market): what other industries need to build things? Are highly fragmented? Require compliance? Are boring?

My first prompt to the LLM:

"Given experience in software engineering and building a product that focuses on construction permitting, and the existence of competitors like X, what is a strategy for building a SaaS product?"

It's not great, but it surfaces actionable insights:

Having a competitor is actually a good sign. It creates a "Second Mover Advantage." ... Your goal shifts from Discovery (finding a problem) to Intelligence (finding a weakness). ... The "Unbundling" Strategy: Look for a competitor who offers a bloated "all-in-one" platform. You can often win by spinning off one of their mediocre features into a standalone, excellent product. ... You cannot compete head-to-head with an incumbent on day one. They have more features and more money. You need a "Wedge"—a narrow entry point where you are undeniably superior. ... Pick a Battleground: Don't try to be "better." Be "better for specific niche."

I'm still interested in energy, is that a viable niche? What industries rely on energy? What industries have strict permitting requirements? Do I go vertical and focus on a single industry, or go broad? This was enough to get me moving again. I need to sit close to problems, focusing on lateral thinking and pattern recognition more than anything else.

Gemini, let me ask you another question:

I am evaluating the feasibility of technology solutions for customer problems in the AEC space. I am looking for information from credible sources that documents known customer pain points. Please provide a report that captures problem areas and potential solutions, grouped by market segment and market cap. Please provide citations.

And another:

What is the market cap of the industry for utility-scale renewable projects? How does it compare to the market cap for small- to medium-sized GCs and subcontractors in the AEC industry? How does it compare to the market cap for renewable solar?

And I talk to friends:

Weird question - do you ever have to worry about permits for your work? Is it all USCG?

And more friends:

Sounds interesting. I've considered doing solar on my roof and the thought of dealing with permitting has been a major blocker.

And Reddit:

Has anyone actually used X? The value prop looks interesting and seems to solve a problem I ran into a few years ago.

Already there are trends emerging. I've narrowed this down to 3 specific verticals, now let's run with it.

What's Next

In the next post, I will share the results of that audit, and then switch gears into teasing out personas and economics of each, reaching out to customers, and grinding on all of the basics