KevinBenSmith

AidfulAI newsletter covering Snipd's new AI Notes feature. The feature provides you with personalized notes from the podcast you listened to, simply by tapping your headphones every time you hear something interesting.

Why should there even be something like cash back or points programs? These programs are taking money from the merchant and giving it to the end-consumer, while the middle-men keep a cut.

I believe a world without these programs (or as you said “watered-down” versions) is more fair for the merchant and others shoppers not using such cards. Especially if the merchant is not allowed to charge the end-consumer with this added fee.

The website wants me to log in…

Haha, love the “Elon vs Marx” one :D

I had similar thoughts about the general concept of using AI to automate AI Safety.

I really like their approach and I think it’s valuable. And in this particular case, they do have a way to score the explainer model. And I think it could be very valuable for various AI Safety issues.

However, I don’t yet see how it can help with the potentially biggest danger where a super intelligent AGI is created that is not aligned with humans. The newly created AGI might be 10x more intelligent than the explainer model. To such an extent that the explainer model is not capable of understanding any tactics deployed by the super intelligent AGI. The same way ants are most probably not capable of explaining the tactics delloyed by humans, even if we gave them a 100 years to figure it out.

As someone who has created several LLM-based applications running in production, my personal experience with langchain has been that it is too high of an abstraction for steps that in the end are actually fairly simple.

And as soon as you want to slightly modify something to better accomodate your use-case, you are trapped in layers & layers of Python boiler plate code and unnecessary abstractions.

Maybe our llm applications haven’t been complex enough to warrent the use of langchain, but if that’s the case, then I wonder how many of such complex applications actually exist today.

-> Anyways, I came away feeling quite let down by the hype.

For my own personal workflow, a more “hackable” architecture would be much more valuable. Totally fine if that means it’s less “general”. As a comparison, I remember the early days of HugginfaceTransformers where they did not try to create a 100% high-level general abstraction on top of every conceivable Neural Network architecture. Instead, each model architecture was somewhat separate from one another, making it much easier to “hack” it.

Pretty much sums up my experience as well.

Congrats on the launch team Readwise! :)

Btw: you need to click on "13min summary" to get to the actual text/audio summary of the episode.

Here's the direct link: https://share.snipd.com/episode/b79b2cf1-b998-4ef9-9ffc-13af...

Can’t access it…

In that case you could have a look at the Snipd podcast app. They have Whisper built in :)

I agree. That’s a great way of putting it.

Great & useful article. We should have more of these for “popular papers”.

Lex Fridman usually asks his guests about their advice for young people. It's one of my favorite aspects of the show. So at some point I decided to collect those sections from the podcast as "snips". I recently went back to them and while doing so, selected my favorite ones. From guests like Elon Musk, Yann LeCun, Demis Hassabis, etc. I hope it's as valuable to you as it has been to me :)

Agree!

This is quite a thorough article by TechCrunch about the podcast app Snipd, and how they use AI to try to recommend you the best highlights in podcasts and let you save interesting moments with transcript.

Love it! :) Have been using it in beta for a while now. It's great to have everything synced automatically. Have used it multiple times to share my highlights of a particular episode with a friend by just sharing the synced Notion page!

We do not yet have AI features for private feeds. But we do have it on the roadmap. Depending on when you tested it last year, you were probably not even able to add your private feeds. In the meantime, you can add your private feeds and listen to them, just like any other podcast app. You can also create highlights, but without the transcript or any other AI features. I've added your vote to this feature request to increase its priority.

We do have this in the back of our minds as a possible way to branch out in the future. But for now, we want to first focus all our energy on creating the best experience for podcasts. I'll add your vote to this feature request ;) I think it would go nicely together, but as you also already mentioned, there are a couple of challenges like how to integrate into an app like Audible.

Completely get that. I always had the same issue with podcasts, which is why I built Snipd together with some friends. If you have the mp3 of the audiobook you can create your own private rss feed with services like SimpleCast and add that to our player. For private feeds we don't yet have our AI features activated, but you can still highlight your favorite moments, add your notes and sync them to Readwise.

Just like all other podcast apps, we are relying on the RSS feed system and are not streaming any audio from our own servers. The podcasts are streamed directly from the podcast hosters via the URLs provided by podcasters in their RSS feeds.

Ok, I added a vote to this feature for you in our internal feature request tracking ;) What do you love about PocketCasts the most?

Cool! That's great. I think there's so much still to do in the space of AI for spoken audio. Be it podcasts, audio books, video calls or similar. So definitely keep at it. I hope you understand that I can't reveal all of our tricks. But what I can say is that it definitely helped us to think about what we as humans use as input for our processing in these situations. So it's not just text. The audio has value as well. To see this you can do a small experiment by just looking at the transcript manually and then trying to find optimal segmentation points. It's difficult. Dynamic ad insertion is also still a challenge for us. We'd like to develop a technical solution for it but haven't found the time to tackle it yet.

We don't have a clear date yet when we will tackle it, but it will definitely come. And the more users request it, the sooner it will come. So I'll add your vot to this request ;) Are your private feeds all podcasts or also some other content?

Haha, I know what you mean. I think almost everyone skips the ads to a certain extent. But you still end up hearing about them at some point. Some of my favorite podcasts are Lex Fridman, Tim Ferris and Andrew Huberman. So there is no way around hearing some "Athletic Greens" ads. And after a while, I actually bought some :D

Great to hear your feedback regarding tipping. Maybe we should prioritize this feature.

Not sure whether this counts, as technically its not CxO, but Elon allegedly changed his title to "Technoking of Tesla"

Thx! With the highlighting, one of our main goals is to make it as easy as possible. At least for me personally, if it's not easy to do, I won't do it. Mainly because I am often on the go when listening to podcasts. And thx to AI, this is now much more possible.

Cool! Nice to hear that you're already using it :) I'll add your vote to the apple watch app request. We do have it on the longer term roadmap. But given that we are still early, we are trying to focus our energy as much as possible on the mobile experience to make sure we get that part right before we branch out. After that, we'd like to build a car, smart watch and web-app version (not necessarily in that order).

Yes, we definitely need to add more info on the homepage. The best way to get an overview of how it works & feels like as a user is this youtube video here, which we created for our PH launch today: https://www.youtube.com/watch?v=jzPekGpC4uw

Let me know if you have any specific questions. Happy to answer :)

Haha, note-taken. This was our first public video and we just recorded it ourselves. Lot of potential for improvement :D

Regarding your questions: Yes, you can import your subscriptions from Castro via OPML. We have a How-To in the app for how to do this specifically for Castro.

The queue & inbox system is still missing, but very high on our roadmap. We're actually working on the queue right now. With the inbox, our plan is to go in a similar direction as Castro has done it.

Let me know if you have any more thoughts on the queue/inbox system.