Píši o budování produktů, AI, letectví a podnikatelské cestě.
80% of Americans with civil legal problems can't afford an attorney. Unbundled legal services is the solution the industry ignored. We built the AI marketplace.
How eight months of research became a working prototype. the technical decisions, the surprising challenges, and the first user reactions to AI-guided legal help.
The legal industry is where healthcare was twenty years ago. Unbundled services powered by AI are the path to making legal help affordable.
Mapping the 2024 legal tech market revealed a crowded enterprise layer, a thin consumer layer, and three critical gaps that AI-powered tools can finally close.
Why I am designing an AI coding assistant that understands entire product portfolios, not just individual files, and the specific problems it needs to solve.
A solo founder's analysis of the legal tech market. who is building what, who is being served, and where the biggest opportunities for AI remain untouched.
The thesis behind building specialized AI agents for law, coding, and beyond, and why general-purpose AI tools leave so much value on the table.
The research phase behind LegalAgento. understanding why 80% of civil legal needs in the US go unmet and how AI-powered unbundled legal services could change that.
ClickAi orchestrates multiple AI agents for structure, copy, design, and SEO to produce coherent websites. Here's how the system works.
Entrepreneurs spent weeks and thousands on websites that could be generated in minutes. Seven years later, ClickAi builds live sites from a voice prompt.
A year of building with AI at the center: integrating Claude across my products and learning what it means to collaborate with AI daily.
What it takes to build AI products that people actually trust in high-stakes, regulated environments like law. transparency, boundaries, and honest uncertainty.
After testing both extensively in production across my products, here is why Claude powers all of my AI features, and the specific technical reasons behind the choice.
After building AI products across multiple industries, here is why I believe vertical AI wins, and why most horizontal AI startups will struggle to survive.
A practical breakdown of what AI agents actually are, how they differ from chatbots and copilots, and why they represent the next major shift in software.
A year of building AI products has given me a collection of expensive mistakes. Here are the five that cost me the most time, money, and sanity.
Building AI for aviation taught me that compliance is not a constraint on innovation. It is a design requirement that shapes better products.
Generating content with AI is easy. Generating content that is consistently good enough to ship is a completely different problem.
Playground prompts fail in production. Here's how production prompt engineering differs - covering consistency, cost, latency, and safety at scale.
ClickAi was already an AI website builder. But integrating LLMs transformed it from a template filler into something that genuinely understands what users want.
Integrating LLMs into production apps is nothing like the demos. Here's what I learned shipping LLM features to real users, mistakes included.
When ChatGPT launched in late 2022, I knew every AI product I had built needed to be rethought. Here is how I approached the rebuild.
Why I decided to rebuild ClickAi from the ground up, what I learned about AI product development, and how v2 became a fundamentally different product.
In 2019, AI for web design meant rule-based systems and template matching. It taught me how to build AI products that deliver value even when the AI is primitive.
I built an AI website generator using rule-based systems and design heuristics. Small businesses need websites and can't afford designers. Here's the origin story.
BorderBot used machine learning to predict US-Mexico border wait times. My first AI product taught me that data products differ from feature products.
