Masters of Scale/Possible: Amjad Masad on vibe coding, AI...

One of the ways you can make models better and more performance is by doing sampling because models are stochastic. You can take a file system, you can fork it a 100 times, run the same prompt with different parameters on these different forks, and then pick the best one.

Masters of Scale · Possible: Amjad Masad on vibe coding, AI agents, and the...

“Every time you run the model, you might get better results, you might get worse results.”

34:21 / 34:54

models better and more performance is by doing sampling because models are stochastic. Every time you run the model, you might get better results, you might get worse results. And if there's a way to verify whether the result is good or bad, you can sample from the model. So what we do is because the system is immutable, has this characteristics of being very cheap to fork. So you can take a file system, you can fork it a 100 times, run the same prompt with with different parameters on on these different, forks, and then pick the best one. Do it repeatedly.

Why this clip

Technical insight that explains how AI performance optimization actually works. The concrete example of forking 100 times provides specific, actionable understanding of AI system architecture.

34:21 - 34:5432sBusiness Mechanics

LinkedIn X

What they said next

There's a role in these organizations that I recently learned about is rev ops - those people that are managing a lot of the data flow within go to market teams. Those people tend to use a lot of SaaS software, but they don't have anything to connect all these things together to create applications and make their salespeople more successful.

11:39 - 28s · Business Mechanics

More from this episode

Masters of Scale

Bold/Contrarian

My advice would apply to 99% of people. I don't think most people who wanna make things with software and computers don't need to learn how to code. It's gonna be a huge cost on them. It's gonna take a lot of time. But instead, just jump into making things.

Direct contrarian stance against the conventional wisdom that everyone needs to learn to code. The 99% statistic makes it quotable and provocative, challenging a widely-held belief in tech circles about coding education.

45:59 - 46:3334s

Masters of Scale

Consequences

There's a woman that had a very rare eye disease and required her to do certain exercises, and she built an app that helps her do these exercises with her eyes every day. There's a woman in Korea and her kid has a very rare disease, and she built an app to manage that disease on a day to day basis.

Powerful human-interest stories showing real-world impact of democratized coding tools. The specificity of medical use cases and personal stakes make it emotionally compelling and shareable.

11:00 - 11:2829s

Masters of Scale

Bold/Contrarian

My generation is the last generation where the tools were really a joy to use, and then it became very industrialized. When we were programming, I started on Infogel Basic. Infogel Basic was a very creative interface. At some point, I think in the mid two thousands, everything became this Linux terminal.

Nostalgic yet contrarian take on programming evolution that challenges the 'progress is always better' narrative. The personal perspective and specific timeline make it relatable to developers who lived through this transition.

7:22 - 8:1452s

Masters of Scale

Business Mechanics

There's a role in these organizations that I recently learned about is rev ops - those people that are managing a lot of the data flow within go to market teams. Those people tend to use a lot of SaaS software, but they don't have anything to connect all these things together to create applications and make their salespeople more successful.

Identifies a specific business pain point (rev ops tooling gaps) that many won't be familiar with. The practical insight about disconnected SaaS tools resonates with anyone in B2B operations.

11:39 - 12:0728s

Masters of Scale

Origin Story

When we were designing Replit Agent, I was playing this video game called Hades. In Hades, every game starts with a different run, and it's randomly generated. It was a very important mindset when designing AI experiences because AI is stochastic. Even the language we use inside the company - an agent run is like a game run.

Unexpected connection between gaming and AI product design that reveals founder thinking. The Hades metaphor for stochastic AI behavior is both creative and technically insightful, showing how inspiration comes from unexpected places.

6:21 - 7:0544s

From the blog

FrameworkOriginal Research

The 5 Clip Archetypes for VC Podcasts

10 min read

Want clips like this for your podcast?

We find your top 5-8 clips, write the hooks, and deliver ready-to-post content. First 2 episodes are free.

Get 2 Episodes Clipped Free

One of the ways you can make models better and more performance is by doing sampling because models are stochastic. You can take a file system, you can fork it a 100 times, run the same prompt with different parameters on these different forks, and then pick the best one.

Transcript

Why this clip

More from this episode

My advice would apply to 99% of people. I don't think most people who wanna make things with software and computers don't need to learn how to code. It's gonna be a huge cost on them. It's gonna take a lot of time. But instead, just jump into making things.

There's a woman that had a very rare eye disease and required her to do certain exercises, and she built an app that helps her do these exercises with her eyes every day. There's a woman in Korea and her kid has a very rare disease, and she built an app to manage that disease on a day to day basis.

From the blog