HN user
Software Engineer @ Palantir
sam@mylaterbase.com
Just to follow-up, I added a basic implementation of bulk uploading (although it's temporarily limited to a small number of uploads until scalability has been given a bit more thought).
Thanks for the feedback!
I'm working on the bulk import as we speak, and it should be ready in an hour or so.
On tagging, do you mean that you would want to define a list of allowed tags that the AI can select from?
Good point, I had planned to add a Mullvad-esque process where you just generate an account number, but didn't want to delay the launch!
If it's any consolation, you don't need to verify your email, so you can just make one up to try it out.
Hey HN, I spent the past few days buiding Laterbase, a (free) AI-powered bookmark manager that enables you to save, search, and chat with your bookmarks.
It's pretty hacky, but would appreciate any feedback!
RIP. I loved reading about Kevin whilst growing up.
Thanks a lot!
Does it need our own OpenAI API key to run this or it doesn't need that?
You don't need an OpenAI API key.
How long are you planning to offer the lifetime offer? If I invest in lifetime plan, is it sustainable enough to run at least for 2 years even if you are not able to sell it more or produce MRR ?
Whilst I'm considering a switch to a subscription model, I will of course honour previously purchased lifetime licenses forever. When you buy a lifetime license, you are also sent a compiled version of the app that you can run locally on your own machine, using your own OpenAI API key, so there's no risk to you even if PitchPal ends up not being sustainable.
Also, how can I contact you if I want to promote this via our platform? Do you have any active affiliate program for promotion?
You can contact me at sam@pitchpal.app - shoot me an email re: the affiliate program as it's not something I've considered!
No, I don't store or read your data. State is managed using your browser's local storage. The caveat being that the tool uses GPT-4, so your data will be sent to OpenAI via an AWS Lambda function.
However, upon purchasing PitchPal you will also be emailed a compiled version of the app that you can run on your local machine, using your own OpenAI API key.
That's not something I've considered, but love the idea!
Initial assumption is that there could be a ton of overlap (e.g. select a research area + provide a brief description of your hypothesis/planned experiments to generate the application).
Are there any nuances / complexities I'm missing? What do you find to be the most time consuming part about writing these applications?
Hey HN, just sharing a tool I made over a long weekend to help founders generate completed applications for popular funds like YC, Seedcamp, etc.
Just enter the name of your startup, along with a brief description of the service + founders, and it will use GPT-4 to create an application that takes into account the requirements of your chosen accelerator.
Yes, I built Markup alongside clinicians within the NHS (they would provide feedback, feature requests, etc.). However, they use a locally hosted version with a standalone sequence-to-sequence model I trained on medical entities; they don't send patient data to third parties.
You can check out some of the clinicians using Markup (+ published papers) in the comment section of my PH release, https://www.producthunt.com/posts/markup-document-annotation
Hey HN! I'm super excited to share Markup with you, which is a totally free & open-source annotation tool that helps you transform unstructured text (e.g. news articles) into structured data that you can use for building, training, or fine-tuning ML models!
Markup uses GPT-4 to learn as you annotate, and speed up the process by predicting complex entities and attributes!
Check it out: https://github.com/samueldobbie/markup
Hey - glad you spotted the thread! RE: your suggestions:
1. Good idea, will be sure to include this in the next release.
2. Sorry to hear that, I've had a few messages about false positives now, and have pushed a temporary fix (https://github.com/samueldobbie/remarkable-extension/pull/4) which causes false negatives instead. I'm thinking the best approach is to mark them in-place as you suggest, with an option to move them to the trash after they've been marked.
3. I'm currently exploring this exact concept! Hopefully, it will update them in the background & enable you to choose a default favicon for pages without one. Definitely seems possible.
Not yet, although I recently made the switch from Linux to macOS, so Safari support is likely to be on the roadmap!
To be honest I haven't tested it on anywhere near that quantity of bookmarks, so I'm not entirely sure how well it'd perform.
My suggestion would be to either wait for the extension to mature a little bit before testing it, or to export your bookmarks first so you have a backup. The extension itself is non-destructive in that it won't delete your bookmarks, but there may well be undiscovered bugs that could mess up your current organization of bookmarks, etc.
Super cool, love it. Is it possible to make a version that plugs into the back of my head?
Good catch, added an MIT license.
The easiest way would be to join this newsletter https://creative-knitter-9009.ck.page/66aa398945 (excuse the domain, just put it together now).
Alternatively, you could subscribe to this issue https://github.com/samueldobbie/remarkable-extension/issues/...
Hadn't really considered bookmark renaming, thanks for the suggestion :)
I like that! Could definitely see it being quite a valuable option, provided it was opt-in.
Planning to add Firefox support soon!
Great idea - appreciate you sharing.
Thanks a lot!
Hey HN, this is just a super simple browser extension that I put together in a few hours to fix the annoying issue of bookmarks becoming outdated (e.g. the bookmarked page being deleted or removed). Just run the extension every once in a while to update the favicons and move dead sites into a trash folder (where you can then decide whether you want to keep or delete them).
Thanks to hrez for the suggestion in this thread: https://news.ycombinator.com/item?id=29796099
--
Repo: https://github.com/samueldobbie/remarkable-extension
Chrome Store: https://chrome.google.com/webstore/detail/remarkable/mjmbcbg...
Strongly agree. Online dating can feel super unnatural & somewhat demeaning compared to organic convos, the typical meat market analogy seems apt.
Funnily enough, I threw together a landing page a couple of days ago [1] to run Google Ads and gauge whether there's wider interest in a solution that would avoid you having to come up with a "hook" + reinvigorate those sparks you mention.
The thought so far is that you'd fill out a pretty invasive character assessment that would be analysed by AI + professional matchmakers to set up one date a month for you with another user (who you won't see or speak to until you actually meet in person).
As in text completions and corrections for software docs? Have been tinkering with this exact concept: https://i.imgur.com/txhgLiy.mp4
What issues do you face with online dating?
That's a good one, the crux of the issue seems to be information asymmetry.
For the time being.
Thanks for the kind words!
Interesting idea, hadn't really considered the standalone concept. Is the desire for a self-hosted portal related to privacy? If so, the extension doesn't store any data, but if you want full assurance, you can self-host the API that's in the repo [1] and install the unpacked extension yourself [2].
[1] https://github.com/samueldobbie/troogl-extension/tree/main/a... [2] https://stackoverflow.com/questions/24577024/install-chrome-...
I love Things and made the switch from Todoist a few days ago. However, I really wish they offered more intuitive and robust natural language recognition of dates / times. Todoists' capture of fairly complex inputs (e.g. "Do X every other day starting Monday") saves quite a bit of time.