anshulbhide

love it!

Also the margins in SaaS used to be 60%; I think for bread its much lower than that.

There's really nothing new in this article - but it's all true.

One major difference between SaaS and bread is the number of varieties in SaaS is much more than that in bread. If you need to customize something for a specific workflow, you can now make your own software than buy something off the shelf and settle for something substandard.

I love the internet.

>>>The core philosophy of MCP is simple: it’s an API abstraction.

That's exactly the problem. As agents become better and can read API documentation themselves, WHY do you need an API abstraction?

Its the ultimate millennial throwback.

> The pull request model on GitHub doesn’t carry enough information to review AI generated code properly — I wish I could see the prompts that led to changes. It’s not just GitHub, it’s also git that is lacking.

Yes! Who is building this?

>SaaS valuations are built on two key assumptions: fast customer growth and high NRR (often exceeding 100%).

They are also on the basis of high gross margins of 80-90%. What happens to margins when you start including token variable costs?

I often summarise HN comments (which are sometimes more insightful than the original article) using an LLM. Total game-changer.

gg anthropic

Can you imagine the insights on human behaviour that she has had?

Yet, this applies for only three industries so far - coding, marketing and customer support.

I don't think applies for general human intelligence - yet.

What happens to Ben's aggregator theory in a world of LLMs where marginal distribution isn't zero any more but the cost of energy and inference?

The majority of these hidden truths are due to senior engineering management in their 40s and 50s who have not coded in decades, and yet pick up the latest trend or fashion and impose that on their teams.

The monolith to microservices trend was one great example of this.

Is SEO basically the same as GEO?

How is it that we always come back to coding in terms of model capabilities?

Agreed - Really surprising this article didn't cover the flip side - how many lives have been saved due to having an instant source of truth in your pocket.

Yeah, stainless steel pipes are used to carry ash slurry away from thermal power plants (especially in China and India). Cast basalt lining increases the lifetime of the pipes by reducing abrasion.

We manufacture cast basalt that's used in applications that have a lot of wear and tear. This is pretty cool to see another application of basalt!

The writing was on the wall when Zuck hired Wang. That combined with LeCun's bearish sentiment on LLMs led to this.

Pretty cool to see an LLM-agnostic memory layer emerging!

Just spent an hour trying to figure out how to create a waterfall chart. ChatGPT's python interpreter failed.

If this works right, this could be a game changer.

Yeah I found it as clear engagement bait - however, it is interesting and helpful in certain cases.

Love this kind of stuff on HN

Really interesting how the AI wave led to major tailwinds for struggling remote gig worker platforms like Invisible and Turing through the data creation and RLHF work.

Benedict Evans caught on early - https://www.ben-evans.com/benedictevans/2023/7/2/working-wit...

This is what HN community is all about. Wholesome posts like this one :)

I wouldn't say he was completely wrong. He was right about "Curiously, the fact that the founders of Twitter have been slow to monetize it may in the long run prove to be an advantage."

Twitter / X punches above its weight (in terms of regular metrics like MAUs and revenue) in terms of cultural impact. One can argue that it was responsible for delivering the 2024 election to Trump. This may have never happened if its original founders had tried to control and monetize it too soon.

> A major project will discover that it has merged a lot of AI-generated code, a fact that may become evident when it becomes clear that the alleged author does not actually understand what the code does. We depend on our developers to contribute their own work and to stand behind it; large language models cannot do that. A project that discovers such code in its repository may face the unpleasant prospect of reverting significant changes.

A lot of companies are going to discover in 2025. Also, a major product company is going to find LLM-generated code that might have been trained on OSS code, and their compliance team is going to throw a fit.