Hacker Newsnew | past | comments | ask | show | jobs | submit | BukhariH's commentslogin

Recently, one of my friends asked ChatGPT about an internal tool with a funny name and ChatGPT seemed to know the meaning behind the name & options to run the tool properly for his usecase.

We googled around to see if there was any information on the web about the tool & there’s nothing on Google which makes sense since it’s a boring internal tool for a financial services company.

Ofcourse it could be a lucky guess or it could be an intern had uploaded the manual to GPT :D


What’s the name of the tool?


It would not be impossible for a LLM to hallucinate a correct answer based on the name of a well named tool that had sane default choices in its architecture


He made the offer at a 40% premium on the April 1st close.

Twitter is a dying social network - sounds like a pretty good deal.


People tend not to be very kind when any product they pay for goes down.

At the end of the day - our companies also have people that rely on our software working in order to do a lot of societal good.


Sure, but it’s incredibly naive to see gh having problems and go “they must not know what they are doing”


It is probably caused from postmortem culture not being shared in the community.

"Having problems" in this world (any kind, not only due to the github scale!) is something that happens - we are not perfect and we work on an incredible amount of layers of complexity.

It is sufficient to actually touch production code on a daily basis to see that it can happen to the best, with the best observability systems or processes. The key is avoiding blaming, and understanding iteratively how to fix the problems underneath (faster recovery, detection time, and so on).


Everybody should be refunded $0.05 for the unavailability of the service they paid for.


Since, Wise & Revolut have been mentioned quite a bit and I'm an ex-Revolut & ex-Wise engineer where I built a lot of the FX infrastructure so I'll throw in my 2c.

My assumption of what you & Neeraj are doing based on your backgrounds: You've decided to build a smart order router for retail FX hence the slow transfers because you need to wait for the spot order to settle T+2.

Story at Wise Funnily enough Wise started in the early days with a fixed fee of £1!

Since all Wise did was match orders between customers so no money was actually moving borders hence no need for conversion or transfer fees etc.

But that fixed fee model was quickly scrapped when Wise started making the market between customers and improving the product experience by introducing rate locks (hedging costs), instant transfers (buffer costs) and actually started considering all the other cogs that go into making an international FX payment.

But there's one significant cost of high value transfers that's been a struggle for both Wise & Revolut to bring down even at scale:

The transaction monitoring, AML & support costs of high value transfers - a single high value transfer can easily cost £3 - £150 for all the due diligence needed to get it done.

If you're going to keep the product fixed fee - this is where you'll need to focus in terms of innovation.

The Story at Revolut Revolut has an Unlimited Zero Mark up FX with a single free SWIFT transfer (every additional international transfer is capped at £5) on the £6.99 premium plan.

But international payments is relatively a recent addition. The vast majority of the time when people are sending money to each other on Revolut, it's a Revolut-to-Revolut or Revolut to local own bank account payment.

Revolut's core strategy is to be the everywhere wallet so that when you're sending money to your family member in Romania from the UK - you just do it as a Revolut to Revolut payment which is bit different to the gap you're trying to fill.

And, there's very little interest at Rev to deal with high value transfers because the AML, transaction monitoring and support costs are too prohibitive when transfers between users are free or capped at £5 for SWIFT payments.

Is there a market for this product? A resounding YES.

This is something we realised at Wise - we weren't as competitive as we liked to be on high value transfers and even though they committed resources to improving this - in typical Wise fashion the progress is very slow.

Now I'm not convinced you'll stick at the £3 mark for long but there's definitely a large untapped market here.

I'm confident that you & Neeraj have what it takes to win in this market & will be rooting for you from the outside!

P.S. The app doesn't seem to be available in the UK app store?


Thanks for sharing your story and for the thoughtful feedback BukhariH! We thought a lot about the current players and offerings in the space, and came to similar conclusions about some of the pitfalls.

That being said, we have made two observations that we are looking to explore by launching Atlantic Money:

1. By deliberately going the super-app route, and trying to be something for everyone, super apps are crappy user experiences, and they aren't a very strong business (outside of south-east Asia).

We are looking to be a focused product, and are committing to not building a bunch of additional features into our app over time that customers probably don't want and are likely going to make the experience adversarial.

2. Super apps are not targeted.

We believe that costs can pile up if you don't have a target customer. Our thesis is that we want to be a very good option for customers moving larger sums of money, and we've explicitly focused on that segment in terms of corridors, and our fixed pricing.

We'll see how it goes with the launch, but would welcome any feedback - either here, or feel free to reach out directly.


How do you plan to become profitable?

£3 flat sounds basically impossible become profitable on. Especially considering you're going for higher volume customers. Yes you have chosen a segment with low overheads (bank to bank transfers), but you will still have the burden of costs that scale very poorly with the volume of £ moved and number of customers served (e.g. CS, AML, KYC etc). Usually revenue (and some costs) that scales with volume is the route to profitability here, but you've explicitly forgone it!

So, in my opinion, your prices will have to rise substantially (most likely) or you will do something akin to what Robinhood did with PFOF and make the customer the product, which is to say, exploit them without them knowing. Sorry if this sounds harsh but your setup currently falls into the category of things that sound too good to be true, would love to hear how you plan to overcome it.

Having said all this I wish you the best of luck, hope to be proven wrong!

(edit: to clarify, i'm considering a business model without significant cross subsidisation - in which case this transfer product would still be loss making in isolation)


Not quite right.

The assets aren’t necessarily stored on OpenSea. You can see how OpenSea gets the metadata like asset url via the docs:

https://docs.opensea.io/docs/metadata-standards

If you scroll down you’ll see that the assets can be hosted on decentralised networks too like IPFS.

Also, not all NFT projects are built the same. I suggest you check out Pak’s Poets project - it’s a very exciting piece of performance art built around game theory.

Disclaimer: I don’t own any NFTs but that doesn’t mean I’m not excited by what they’re doing.


I agree, you can store it on IPFS. But most people don't, they don't even know it's possible.

I mean, most people could host their own video, but they upload on youtube, and just youtube.

Well, this is the same, only even more niche, so the easy "upload" on open sea is very, very attractive.

I'd wager if one the most popular NFT trading plateform disapear, 99% of all NFT would go poof.


The majority of all NFTs are stored in IPFS; source: I run a platform indexing NFT media files.

Also, the art is the token. Data storage is a orthogonal problem entirely.


> the art is the token

The art is people believing they own more than that.


> The majority of all NFTs are stored in IPFS

Oh, I would never have believed it. Thanks for the info. Most people that I know don't even know what IPFS is. Are those human uploaded, or bot uploaded?

> the art is the token

Yes, but why isn't the token displayed in the centered of the page on most popular platforms?

Because human needs a symbol that represent that token.

For most people, the animation and the number are the same. NFT work because people share a believe about them, after all.

And all the strong believe in humanity are sustained using symbols.

BTW, what's the name of your platform?


I was surprised you could just store the NFT payload in a Dropbox folder... and also swap out the contents whenever you want.

Buyer beware, I guess


“The sector is hot hot hot — many more people want to invest and there really is no product to offer real free investing in Europe just yet. (While we acknowledge that there will be a few choices), I believe the key is to be more European here to win, that means offering sustainable investing choices (vs incentivising people with too much leverage, etc).”

-- Taavet Hinrikus


In London these dark/cloud kitchens are becoming more & more common.

You might have already been ordering from them & not realised - they used to be marked as “Editions” on Deliveroo:

https://foodscene.deliveroo.co.uk/promotions/deliveroo-editi...


ZM has two classes of Common Stock - Class A & Class B.

Looks like Eric exited his Class A shares but he still holds over $4B+ USD in Class B.

source: https://docoh.com/company/1773298/yuan-eric-s/insider-owners...


He didn't exit class A, he didn't hold it in the first place. Instead, he converted a small percentage of his class B into class A, and sold it.


Converting from Class B -> Class A opens a position in Class A.

Selling those Class A shares is still exiting a position in Class A.

But you're definitely right about the context!

He merely converted some Class B shares (non-publicly traded with 10X voting rights) to Class A (publicly traded) in order to sell them.


Currently working on my French - was super interested to use this but it's not working.

Looks like the request to:

https://www.cliplingo.com/lesson/start?id=18

Returns an empty video id: https://pastebin.com/8JemA2T3

Hopefully you can fix it soon - super excited to use it!


Thanks for heads up! It is fixed now.


That's a very similar pattern to what I've been using at TW.

At the core of our implementation is this library to schedule the jobs to different workers: https://github.com/kagkarlsson/db-scheduler

Would highly recommend that library!


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

Search: