The New Standard for Custom Software: AI-Assisted Development
We've reached a turning point in software development. Building custom software the traditional way — line by line, by hand — is effectively a thing of the past. AI language models (LLMs) now represent the most sensible approach to creating custom software, and we are fully committed to that future.
Why the Shift Has Happened
- AI-generated code quality has improved dramatically in a very short period.
- Both frontier models and locally hosted models continue to get better, faster, and more capable.
- The fundamentals of programming are largely a solved problem — the focus is now on what to build and how to specify it, not on manually writing every line.
What This Means for You
Increasingly, technically oriented individuals and teams can build their own software without hiring a development firm. We want to be straightforward about what that still requires:
- Time. Meaningful applications don't appear instantly.
- Expertise. Non-trivial software demands real knowledge of what you're building and why.
- A proper plan. A single prompt can go a long way, but complex software requires a formal specification and a step-by-step plan before any code is written.
- Thorough testing. AI-written code still must be rigorously verified.
- Human review. Every codebase should be evaluated — either manually by an experienced reviewer or with the help of advanced models, which have become remarkably effective at identifying bugs and vulnerabilities.
What We're Doing About It
The AI-driven future of software is both exciting and unsettling — and it is here to stay. Career developers have no choice but to embrace it; we've chosen to lead with it.
We are currently putting together a delivery plan that leverages AI to help you get the custom software you need — at a reduced, fair price — without you having to become an expert in AI tooling yourself. We'll be sharing details of this plan soon.