Why you can’t fire your engineering team just yet
Natural language fails to manage complexity and engineering requires precise mathematical logic.
ByteSauna Blog
The latest posts in the ByteSauna blog.
Latest posts
AI Enables You To Do What You Already Can Do
AI is a talent multiplier, not a substitute.
The vision of combating model hallucinations with formal verification
Explore the shift to AI-driven coding, the trust gap in GenAI, and how formal verification via Lean can replace manual review with mathematical certainty.
Execution over inspiration
Ideas are overrated; execution is everything. Sustainable routines beat genius.
The synthesist
A grounded take on AI as an abstraction layer, why jobs persist, and how future opportunities favor synthesists who connect engineering, business, and context.
Where 2025’s agentic AI hype fell short
Why AI agents missed 2025 hype: failed pilots, slower developers, and a rethink of LLMs, what AI really is, why assumptions misled companies.
Language is the new UI
LLMs aren’t minds or agents—they’re interfaces. An argument for viewing AI as an abstraction layer, not artificial intelligence.
On insight debt
Insight debt is the hidden tradeoff of AI-generated code: faster output today, harder maintenance and lost flexibility tomorrow.
What is creativity?
Is AI creativity real or just imitation? A reflection on human uniqueness, artistic style, humility, and whether true creativity depends on subjective experience.
Understanding mathematics through Lean
An introduction to how Lean formalizes mathematical proofs using the Curry-Howard correspondence, turning theorems into types and proofs into verifiable terms.
Microservices should form a polytree
Microservices work best as a polytree with clear dependencies and no cycles. Learn how this structure reduces complexity and makes systems easier to scale and maintain.
Do the thinking models actually think?
LLMs mimic understanding but think bottom-up, unlike humans. Explore why they’re more than autocomplete and why the future is human–AI collaboration, not replacement.
No free lunch in vibe coding
AI won’t replace coding, it will simply move complexity from code to prompts. True control and precision will always need technical skill and structured thought.
My favorite math problem
There’s one math problem I find surprisingly deep in how it connects simple puzzles to profound ideas in logic, proof, and the rise of AI in mathematics.
LLMs are steroids for your Dunning-Kruger
ChatGPT and LLMs amplify both insight and illusion. They are not knowledge engines, but confidence engines shaping how we think and learn.
I love boring business ideas
Boring business ideas often outperform flashy startups. Stability, predictable revenue, and solid foundations create sustainable success without heavy funding.
AI can code, but it can't build software
AI can write code, but it can’t build real software. Software engineering remains human work because AI can code, not engineer.
Don't Let an AI Ruin Your Voice
AI can multiply your output overnight... or quietly erode your brand. How to keep efficiency and authenticity in balance?