Roman Dobrik

30 years of experience in various product development roles in leading Silicon Valley enterprise technology companies. Cofounder of Collaxa, acquired by Oracle.

[Roman Dobrik @ScaletonOne]

@Cernovich Actually, Israel brought millions of highly educated people from former Soviet Union countries, and they contributed significantly to Israel's technology miracle.

[Roman Dobrik @ScaletonOne]

@tunguz But they must talk FTP. Ok SFTP. Fax in Germany

[Roman Dobrik @ScaletonOne]

@brockpierson Yes. Netscape is still an underappreciated company. The entire internet stack we use today owes to it. Unfortunately, also JavaScript 😕

[Roman Dobrik @ScaletonOne]

@Clint_Davey1 Also they are responsible for bad Mexican music 😃

[Roman Dobrik @ScaletonOne]

@Clint_Davey1 Czech not Austrian

[Roman Dobrik @ScaletonOne]

@github @code Finally, I was able to work from my gym, running dev agents and checking the status of my tasks between sets 🙂

[Roman Dobrik @ScaletonOne]

@tunguz

[Roman Dobrik @ScaletonOne]

We will see where the market will head after those IPOs. While Google or Facebook built solid, well-integrated, and controlled highly profitable pipelines and increased shareholder value, most of the crypto value simply evaporated. For AI companies, revenue will be there, but what about the margins? They will have to share it with other layers: hyperscalers, hardware, energy. Also, while Google and Facebook founders maintained control over those companies and made difficult decisions faster, both OpenAI and Anthropic have a much more diverse shareholder structure with many often competing interests.

[Roman Dobrik @ScaletonOne]

This is nothing new under the sun. When I was working at Netscape almost three decades ago, we did a lot of work on corporate sites, banks, insurers, and telcos to bring them to the internet. You have to deeply understand their infrastructure and business processes to make any groundbreaking technology useful in real-world scenarios.

[Roman Dobrik @ScaletonOne]

@deedydas

[Roman Dobrik @ScaletonOne]

I've been in Silicon Valley for some time, so it's nothing new. With dot-coms, Google, or Facebook IPOs, there was always a feeling of missed opportunity. But this paranoia is also driving innovation here. As the godfather of Silicon Valley, Andy Grove, said, “Only the paranoid survive.”

[Roman Dobrik @ScaletonOne]

It was beneficial for all parties. I would say net profit for old members was actually higher. German, French, and Italian companies were able to acquire assets for peanuts. And let’s not pretend that all citizens of old members did not benefit. Increased profits from corporates transferred to social benefits. Without being able to lower costs of production, west European corporations would not have been able to compete a long time ago. Well, today when the cost of production in eastern EU members is starting to match, European companies are starting to lose competitiveness. Not counting catastrophic energy policies and extreme overregulation.

[Roman Dobrik @ScaletonOne]

@anshublog Well, not necessarily. Special interest groups satisfied, problem solved.

[Roman Dobrik @ScaletonOne]

@anshublog Maybe we can return blockchain to its original purpose: a tamper-proof persistence layer for sharing data and context between agents. It can also serve really well in cross-organizational agent communication, which is the biggest weakness of existing agent platforms.

[Roman Dobrik @ScaletonOne]

@anshublog Working on something like that. Like Collaxa 2.0 run by an agent. Persistence is the key for any agent-driven business process.

[Roman Dobrik @ScaletonOne]

Yes, we agree. However, the number of lines of code generated should not be considered an important metric of quality for AI coding agents. Similarly, a junior developer who writes thousands of lines of code is not more valuable than a senior one who can fix 10 important lines and more importantly design a high-quality product.

[Roman Dobrik @ScaletonOne]

Scalability is the new MVP—actually, it always was. But today it is even more important, as the time from idea to mass deployment has shortened from months to days. Projects must also be able to scale vertically, enabling customization. That is really hard without a well-designed modular architecture, not to mention code maintenance. I was never a big fan of MCP, as it usually didn’t solve the hard problems first. But it was able to secure financing.

[Roman Dobrik @ScaletonOne]

@RiturajBiswas @garrytan Well, you probably don’t need to write 100,000 lines of code per day for an MVP. I don’t think Garry is interested in companies that don’t scale to millions of users. 🙂