Hacker News new | past | comments | ask | show | jobs | submit | mvkel's comments login

OpenAI has been a product company ever since ChatGPT launched.

Their value is firmly rooted in how they wrap ux around models.


this is actually a helpful little utility for someone applying to yc, or trying to raise.

when refining a pitch, it takes a lot of, well, pitching to discover what's wrong with your message, or where it's falling flat.

the more you can simulate this pushback from different angles, the better your pitch will be when you eventually present it.


You ask it for negative feedback, and it generates a negative feedback text. There's nothing helpful there — it's a catastrophizer's Swiss Army knife.

Solid businesses with a real chance at an upside don't need much convincing or fielding criticism. If you have a profitable company that needs to scale to do more business, just walk into a bank... If it's a pie-in-the-sky app idea that could maybe be monetized after market capture, then I don't think anyone needs imagined negative feedback at that point.

It's good satire, though.


Most founders don't have access to experienced VCs or YC partners who can poke holes in their pitch before the real thing. Even solid businesses with good fundamentals can struggle to communicate their value effectively.

Sure, if you're an established profitable business seeking straightforward financing, you might not need this. But most YC applicants are early-stage founders trying to articulate vision, potential, and execution capability. This tool might be able to refine that message.


I'm not sure what your point is. If you already have a successful co that doesn't need to raise, then yeah you probably don't need to practice seed stage pitching.

Also:

> If you have a profitable company that needs to scale to do more business, just walk into a bank.

Where they'll ask you to put up all your personal assets as collateral in exchange for a 20% loan on your existing accounts receivable


My point is that narrative-driven funding is very often the wrong approach. Before people polish their storytelling and pitch decks, they should look at the fundamentals of their business. Then, they probably won't need the whole startup theater to try to get funding. Moreover, if the fundamentals are so bad they can't be funded, it's a signal about the business, isn't it? Maybe that doesn't apply to some one-off unicorn businesses, but it applies to the 99%.

I'd say people are better off going with fundamentals. Because if your fundamentals are bad, but let's say you manage to enthral the investors with your confidence, charisma, and selective optimism, they may fund you, but your negotiating position will be incredibly weak. If they stagger the funding over milestones, upsetting your investors for any matter - product related or not - will mean game over. If they want a lot of equity, you will have to say yes, as they'll still be the one investor tentatively interested, etc. It's a bad deal. It's risking a lot of structural damage to a new business.

With good fundamentals, all of that doesn't apply. If you start a company, get some cash flow going, and show profit, investors will start chasing you. Whatever funding you would have gotten through the usual Silicon Valley venture pitching, you will now get on actually good terms.

> they'll ask you to put up all your personal assets as collateral in exchange for a 20% loan on your existing accounts receivable

Has this happened to you? My experience is that banks offer almost up to 1:1 debt to (company) assets, so long as the business model is clear and profitable. If someone's trying to fund an unprofitable business (like I said, for market capture, for example), then there will be problems. But in that case, the fundamentals aren't great.


> Turns out their ML/AI method doesn't generalize so well.

I'd argue the opposite. AI typically generalizes very well. What it can't do well is specifics. It can't do the same thing over and over and follow every detail.

That's what's surprised me about so many of these startups. They're looking at it from the bottom-up, something ai is uniquely bad at.


I really wish people would stop evaluating a model's coding capability with one-shots.

The vast majority of coding energy is what comes next.

Even today, sonnet-3.5 is still the best "existing code base" model. Which is gratifying (to Anthropic) and/or alarming to everyone else


This works for blog posts, certainly. But it falls apart if you're doing anything even slightly long form, or have multiple points to make.

It's also why LinkedIn posts all sound the same.

"It seemed like any other Monday. Little did I know, it was going to be the day that changed my life forever..."

"Marketing isn't about getting the most traffic. It's about converting the most traffic. A thread:"


Why do you think the inverted pyramid doesn't work for longer form?

If you have multiple points that don't both support a larger point, they should probably be split into two separate essays.

Your first example could be the start of an inverted pyramid if the thesis of the post is how the Monday was just like any other. But the next sentence dashes that notion.

The second example could be an example if it quickly follows up with the ways to convert traffic, but better to lead with the novel way(s) to convert traffic, then follow up with why conversion is more important than generation.


To :+1: this, even if it's a book - there is a central thesis - a headline and a sentence that tells you whether you want to read more. "Your pet could save your life" - The six surprising reasons that people with pets live longer than others.

Then each chapter has the same: "Getting in touch" - why stroking your cat soothes your body. Etc

You may even have sections within the chapters and each can follow the same format.

Thousands of years ago it was enough just to write down stuff you've learned, call it "Meditations" and hope people would still be reading it in the distant future.

Now if it's just "stuff I've learned about coding" or "things that make me happy" you're going to need an extremely strong hook to tie that together and build an audience.

So start with a single thesis and decompose from there. Inverted pyramids all the way down :)


This sounds like a business book on cats. Useful, yes, but not something I'd read for its writing value.


To me it just gets repetitive. After the first one, my brain recognizes the pattern. If chapter 1 starts with a bang, then fills in the blanks, then chapter 2 is structured in the exact same way, it feels formulaic; not good writing


The two examples you gave seem like bad examples of inverted pyramid. Inverted pyramid doesn't leave you hanging. It's not clickbait. It should be the case that within 1-2 sentences you can mostly understand what the rest of the article is going to be about (like an abstract).


It doesn't work for me, I get a little angry each time I read the above-the-fold five word hot take.


I'm sure they're still deploying years of loss carry-forwards. Nothing wrong with that, but I doubt they're generating tax revenue yet.

That'll eventually change of course, and they'll be the best space game in town.


The 4-star system is indeed wonderful. A rating that is truly "neutral" doesn't exist in a world of human subjectivity


I think this release is for the researchers who worked on it and would quit if it never saw daylight


That looks... out there. Is it parked at a Lagrange point?


Article mentions that it's in a highly elliptical orbit.


Very strange. Almost like the altimeter setting wasn't set to the right number?


Doubtful. The radalt should have been screaming "sink rate" or something similar in the cockpit regardless of the pressure setting.


Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: