With the Help of AI, Founders Are Building the Tools They Always Wished They Had
By now, we have become accustomed to the idea that artificial intelligence can turn a concept into a thing. Inevitably, some of what emerges is pointless, silly, or purely experimental (an accomplished musician friend of mine uses the AI music generator Suno to write catchy songs with nonsensical lyrics–one is bafflingly titled “Big Spaghetti Driver”). Other people use AI to build a concrete solution to an everyday problem that has been staring them in the face but was, before now, too time-consuming and expensive to tackle.
At AllBusiness, we’re interested in entrepreneurs who are doing just that: using AI to find or create tools to make their lives–or those of others–easier. One such entrepreneur founded the company GovKodo after realizing that local governments needed systems that were easier for both government employees and residents.
Julie Stipe: Can you briefly describe what GovKodo is and what it does?
Luke Peters: GovKodo is a web-based software company with the main functionality of providing online permitting for local governments. It allows citizens to apply and pay for building, zoning, and numerous other permits digitally through an online portal without needing to print, fax, or handwrite and deliver documents. It also gives local government staff the ability to conduct plan reviews and organize and issue permits virtually.
The purpose is to massively increase convenience, reduce friction at every step of the permitting process, reduce cost and review times, and provide a central repository for all permits. GovKodo also provides a suite of related services, such as an inspections and code compliance mobile app, permit parcel map, and utilities work orders and asset management.
Stipe: Tell us about what led you to start GovKodo. Were there specific frustrations in your government work that contributed?
Peters: The genesis of GovKodo was almost entirely because I needed what I thought was an extremely simple feature from my locality’s current permitting vendor and was told no. Well, maybe I have oppositional defiant disorder or maybe I just want to uncover how everyday things in this world actually work, but I decided to challenge that no.
I’ve always wanted to know how difficult it is to create a website that users have to sign into from scratch, and found out quickly that nowadays there are existing free tools that do the heavy lifting for you. Once I started clearing hurdles that I had previously considered exclusive to the ramparted realm of genius software engineers, my confidence grew, I started adding bigger and more important features, and I finally said, “I can do this. I can do this better than they can,” and pulled the trigger on the LLC and other subscriptions and business necessities.
Stipe: What’s one task your software can perform that used to require much more time or many more steps?
Peters: If you mean the leap from in-person paper processing to a fully online and automated system, we’re talking thousands of person-hours of work and untold gallons of gasoline saved by giving citizens the option to complete applications online as opposed to driving across a county that’s however big and hand-delivering all their application materials (which may not be complete, might need revisions, etc.).
As for GovKodo versus other software solutions, it’s more about the unfettered customization it allows the end users. That doesn’t just mean the settings and customization within the app itself, but being able to provide crucial, individually tailored features on the fly. Does your locality have a special rule for permits? Do quirks in your workflow demand that a section be more flexible? I can cook that up for you within a week, and without charging you anything extra, because your request simply became part of the weekly upgrades that add useful functionality to the app for everyone.
With legacy companies, you’d have to write up a new contract and expect an invoice for a thousand dollars just to add a button that should have been part of the core functionality for all customers to begin with. GovKodo, on the other hand, lets administrators control and request everything they want. Maybe we’ll run into a bottleneck with requested features and have to triage at some point, but it hasn’t happened yet.
Stipe: How did AI help you build GovKodo? What tools or platforms did you use?
Peters: AI is everything. I had taught myself a little Python before, earned some edX certificates, taught myself how to make web scrapers and games, and picked up some HTML and CSS back in middle school—but never would I be able to produce this product in its entirety and be able to iterate and upgrade so quickly without LLMs. Unlike writing, which they are still terrible at, coding is the perfect use case for language models, because there are such well-defined rules and documentation. Of course, having the base coding knowledge lets me ask intelligent questions, guide the AI towards better solutions, think about the system as a whole, and be able to troubleshoot small things when it inevitably messes up. I don’t think you can dive in 100% naively.
I’ve been using Virtual Studio with the integrated Copilot chat, which lets you pick which LLM model you want to use (Claude, ChatGPT, Gemini, etc.) while supplementing with Cursor to double my output and make them adversarially check each other’s work. There are some tools that let you hand over the keys to your code and the AI will spend 9 hours overnight going nuts on building whatever you want, but I’ve found it useful to review the code and ask questions after every little sprint.
Stipe: Have you seen an increase in AI use by local government over the years? How do you see local governments incorporating AI use into their operations in the future?
Peters: I don’t know if we’ll see local government functionally replaced by AI. Maybe we’ll see it in customer service-related roles, and for staff to use in their spreadsheets, research, and possibly plan reviews. Actuarial tasks will get easier. But many functions of local government necessitate boots on the ground and human interaction and review. However, local governments will certainly employ AI-powered vendors when there’s a technological need, in the form of drones and GIS, sophisticated utilities and infrastructure inspections, permitting software, etc.
Stipe: Are there other outdated systems that you think AI should be used to help overhaul?
Peters: It’s kind of wild that you can instantly give any reasonably intelligent person the power of 5 junior coders, and the tech we interact with is still so bad. Employees at these large corporations are paid handsomely and are supposedly doing all this stressful work, and what do users get? Terrible UIs (worse every update), lag, fewer features, and glitches. Clearly, the revenue is being made elsewhere.
You would assume that the legacy software dominating market share has cemented itself in the market because these companies have solved all the issues and have an airtight system. But then you go to use the software and it’s broken, ugly, and almost unusable.
I think there’s a lot of institutional backlash to change, across numerous industries (local government not least among them). Old dogs do not even want to consider learning new tricks. There are certainly some real migration costs, and I also tend to live by “better the devil you know,” but I think human sentiment is keeping a lot of organizations from moving forward more quickly.
Stipe: What’s one thing other founders using AI in their startups should keep in mind?
Peters: You do not have a moat. Anyone can make what you’re making. All you have is speed, branding, first mover advantage, and—if you’re lucky—some degree of institutional capture. Instead, use the advantages AI gives you to be radically responsive to your customers.
Stipe: Do you think AI will lead to a boom in startups?
Peters: Much smarter people than me are already carving out niches in game-changing sectors of everyday life. Home security, manufacturing, procurement, vision and robotics, everything. These are becoming cottage industries that used to be walled behind expensive investments.
The only questions are how saturated the market will get and whether we’ll see incremental improvements from competition and innovation or whether it’ll be a race to the bottom by people trying to undercut one another. And then it’s only a question of when the big players in AI pull the rug out from under their current widely distributed userbase model, leaving the power of AI tools to those who can afford to pay the equivalent of software-dev salaries for them.
Stipe: What advice would you give to founders trying to solve problems in slow-moving industries?
Peters: Don’t be scared to try to fix the problem yourself. If you don’t, no one will.
