# We should be more ambitious with the software we build

> Software builders can now create complex tools in hours instead of years with modern AI capabilities.
- Title: We should be more ambitious with the software we build
- Summary: Software builders can now create complex tools in hours instead of years with modern AI capabilities. Software developers can achieve ambitious projects in…
- Keywords: software, coding, tooling, openai, rust, pgrust, postgres, technology, startups, more, ambitious, build
- Source: Danluu — https://danluu.com/ambitious
- Read time: 1 min
- Topics: software, ai, coding, tooling, openai, rust, pgrust, postgres, technology, startups
## Malis shows pgrust testing
Michael Malis tests pgrust thoroughly uncovering real bugs in Postgres including numerical, JSONPath, and memory safety issues.

> This numerical bug that can cause valid SP-GiST operations to fail
## pgrust performance focus
pgrust's main pitch is performance improvements that could benefit Postgres and Clickhouse if implemented.

> The main pitch for pgrust is actually performance
## Vibeslop projects skepticism
Most people dismiss rewrite projects as vibeslop garbage because they lack time to verify quality.

> If I didn't know Malis, I would make the same assumption
## Software speed improvements
Rust compilation and filesystem overhead can be sped up significantly with modern AI capabilities.

> Ripgrep, compiling rust, syscall and filesystem overhead
## Regex engine development
Author built a regex engine faster than general-purpose libraries for specific workloads in hours.

> I built this regex engine that can be tweaked to be faster than other general-purpose regex libraries
## Rust interpreter project
Author created a Rust interpreter that speeds up cold builds by 50x with minimal effort.

> I built a rust interpreter that gets a 50x speedup on cold compiles
## Board game AI development
Author achieved superhuman board game AI in 5 hours with minimal AI knowledge.

> It took about 5 hours of my time to get something superhuman
## AI tooling examples
Author built multiple tools including a regex engine, rust interpreter, and board game ai in hours.

> For syscalls and filesystem overhead, just switching development from Mac to Linux improved this a lot
## Postgres bug fixes
pgrust uncovered real bugs in Postgres including numerical and memory safety issues.

> This money overflow bug
## AI model capabilities
Modern AI models can solve complex problems that previously required years of work.

> I can do a lot more with modern ai capabilities
## Rust compilation speed
Rust compilation time has been reduced significantly with modern AI tools.

> Rust compilation is slow but can be sped up with ai
## AI tooling impact
AI tools have made software development faster and more ambitious for individual developers.

> This set of projects would've taken a year or more to get to works for me quality
## pgrust verification
Malis uses Kani and Lean for code verification in pgrust projects.

> He's since switched to lean and has been using lean to verify things that are too complex to do in kani
## AI model benchmarks
GPT-6 Astra passes zstd tests with 99% accuracy across all conditions.

> 6 Astra succeeds at passing all of the registered tests more than 99% of the time
## AI tooling examples
Author built tools for specific use cases that would've taken years previously.

> For example, the last time I taked to him about pgrust verification
## AI tooling impact
AI tools have made software development faster and more ambitious for individual developers.

> This set of projects would've taken a year or more to get to works for me quality
## Key takeaway

Software developers can achieve ambitious projects in hours rather than years using current AI tools.