ST-DOS
sininenankka.dy.fi 78 9Writing Gnome Apps with Swift
swift.org 377 143BeagleY-AI: a 4 TOPS-capable $70 board from Beagleboard
beagleboard.org 75 34Launch HN: PointOne (YC W24) – Automated time tracking for lawyers
Hi HN!
We’re Adrian, Katon, and Jeremy from PointOne (https://pointone.ai). We’re building an app that automatically figures out what lawyers are doing and generates timesheets for them. Here’s a quick demo: https://youtu.be/yrL3e1hgaNc, and here’s an even quicker one: https://youtu.be/giIaAxZp2M0.
If you’ve ever hired a lawyer, you know most of them bill by the hour—or more specifically, by the 0.1 of an hour (hence our name!). What most clients don’t realize is how painful it is for lawyers to track all their work in 6-minute increments. Lawyers hate time tracking, and many say it’s the worst part of their jobs.
Adrian started out his career as a corporate lawyer at Fenwick & West. The first thing he was taught was how to track and bill his time. Between the 70-hour work weeks and billing to 10-15+ clients per day, staying on top of timesheets is surprisingly hard. To make things worse, law firms are extremely particular about how narratives (that is, descriptions of tasks performed) are crafted—down to the punctuation and diction required. So, Adrian became chronically delinquent in submitting his timesheets, and the firm threatened to take away his bonus multiple times as a result.
Attorney time tracking is not a new problem, and companies have been promising to solve it for years. But pre-LLM attempts at automatic timekeeping never worked as advertised. We were inspired by products like Rewind, and felt that a narrower vertical application could finally solve this problem for lawyers.
Our product is a desktop application that a lawyer turns on at the start of their work day. It runs passively in the background and captures logs from everywhere they work: the OS itself, Word, Excel, calendar, emails, web browser, Slack/Teams, etc. We then clean, pre-process, and interpret the logs. Modern LLMs enable a bunch of cool features. For example, we can pull subtle context from an attorney’s browser activity to associate that work with a client. And for each client and project, we use these models to generate a time entry with a narrative description that matches both the firm’s and the client’s style preferences.
Besides the fact that lawyers hate timekeeping, using PointOne lets them be sure that they’re not letting time slip through the cracks, and frees up hours per week they can spend on other things. It also helps firm leadership by getting more consistent narratives, and faster timesheet submission.
Given the sensitivity of the data captured, privacy and security are massively important. As such, we have customizable data retention periods, we do not use firm data to update models, and we encrypt all data (in addition to employing other standard practices for processing confidential data).
Since our app primarily works for legal workflows, it might not be super useful for most people here (maybe some though!). We would love it if you could check out our demo video, leave your thoughts in the comments, and introduce us to any lawyers you know.
Peter principle
en.wikipedia.org 115 62Show HN: I built an interactive plotter art exhibit for SIGGRAPH
I'm enthralled with using pen plotters to make generative art. Last August at SIGGRAPH, I built an interactive experience for others to see how code can be used to make visual art. The linked blog post is my trials and tribulations of linking a MIDI controller to one of these algorithms and sending its output to a plotter, so that people may witness the end-to-end experience.
Misunderstanding about the details of how Apply Pay works
birchtree.me 211 193DBRX: A new open LLM
databricks.com 714 296The last crimes of Caravaggio
newstatesman.com 132 20Show HN: A (marginally) useful x86-64 ELF executable in 466 bytes
github.com 72 16Things I Learned from René Girard
honest-broker.com 45 21Launch HN: Patchwork (YC W24) – Team communication based on feeds, not chat
Hey HN! We’re Nikki and Dhruhin, co-founders of Patchwork. We’re building a communication tool for teams that lets you stay in sync while at the same time staying in flow. It’s centered around a ranked feed instead of chat.
Edit: as requested, here are some screenshots: - Feed: https://imgur.com/a/bvH7ypQ - Post Creation: https://imgur.com/a/HENe15A, Chat: https://imgur.com/a/MVyVykY.
Over the last several years, we’ve noticed that it’s getting increasingly difficult to stay in flow. We believe it’s because chat (i.e. Slack, Teams, and similar tools) has evolved into something it wasn’t designed for. Chat originally served as a way to free us from our desk by giving us the safety that if we were needed for immediate matters, people could reach us. Now it’s become a dumping ground for all communication: daily updates, product and engineering discussions, announcements, etc. Both of us still reminisce about the days of in-office work, before Slack became mainstream, where everyone abided by the headphone rule—the unwritten pact that headphones meant someone was in deep work and not to bother them unless you really needed to. Compared to then, the onus has now shifted to be on us to determine which chats are urgent and should take us out of flow vs. which messages can be responded to later. It feels like the very tool that was meant to liberate us instead made us beholden to its pings.
Patchwork is our attempt to solve this problem by shifting the primary communication model from group chats to feeds. Posts are made in relevant groups and each team member has a home feed personalized to them.
The feed algorithm evaluates each post's relevance and urgency based on a bunch of factors, including the post’s content, user's role, ongoing projects, and recent interactions. Our goal is to maintain a high signal-to-noise ratio with our feed’s algorithm so we can surface the most important information first. When you’re not in flow or in between meetings, you can check the feed to stay in sync.
There have been feed-based work communication products before, but they’ve often overlooked the fact that writing a post has more friction than writing a chat message, which is why people often revert back to doing everything over chat. We’re combating this by using LLMs to create a better writing experience (ie. generate title and tl;dr, simplify selected content, change the tone, etc.).
As a product team ourselves, we know that much of our work happens on different platforms. We’re building integrations with the likes of Github, Linear, Figma, GSuite etc. Having all of this activity from different platforms also rolling into our feeds allows us to stay in sync with all the different work being done on our team instead of having to check various sources of data.
Lastly, we do have chat on the platform, but chat looks and feels like chat. It’s meant to be used for immediate needs.
Here’s a demo of the product: https://www.youtube.com/watch?v=SA3rmSjNjDw
Since it’s a team communication product, it's hard to use it in single player mode, so we don’t have a “try it now” link to jump straight in. But you’re welcome to email us at hackernews@atpatchwork.com and we’ll onboard your team.
The product is still early with a basic feed, a few integrations, and simple LLM assisted post creation, but the main flow is already there, so if the message resonates with you, we’d love for you to give it a try.
More importantly, we’d love to hear your ideas about team communication and getting it back to working for people, not at them!
https://www.atpatchwork.com/
Infinite Mac: Turning to the dark side
blog.persistent.info 200 43Cliff Stoll, the mad scientist who wrote the book on how to hunt hackers (2019)
wired.com 131 71The Pentagon's Silicon Valley Problem
harpers.org 208 273Want to start a startup? Meet all the YC partners in Boston – Apr 20th
ycombinator.com 99 75A step beyond Rust's pattern matching
radiki.dev 52 8The window for great-grandmothers is closing
memoirsandrambles.substack.com 206 324We've been here before: AI promised humanlike machines – in 1958
theconversation.com 15 7AI hallucinates software packages and devs download them
theregister.com 8 1FuryGpu – Custom PCIe FPGA GPU
furygpu.com 343 106Finley (YC W21) is hiring to remake the $1T private credit space
finleycms.com 0 0Astronomers discover a rare eclipsing X-ray binary
phys.org 10 0Daniel Kahneman has died
washingtonpost.com 869 260Recent 'MFA Bombing' Attacks Targeting Apple Users
krebsonsecurity.com 355 206NASA's Europa Clipper Survives and Thrives in 'Outer Space' on Earth
jpl.nasa.gov 53 5MTA board votes to approve new $15 toll to drive into Manhattan
nytimes.com 339 659Digital signs in Brookline are collecting data from your phone as you walk by
brookline.news 78 58A review of zero-day in-the-wild exploits in 2023
blog.google 16 2Launch HN: Aqua Voice (YC W24) – Voice-driven text editor
Hey HN! We’re Jack and Finn from Aqua Voice (https://withaqua.com/). Aqua is a voice-native document editor that combines reliable dictation and natural language commands, letting you say things like: “make this a list” or “it’s Erin with an E” or “add an inline citation here for page 86 of this book”. Here is a demo: https://youtu.be/qwSAKg1YafM.
Finn, who is big-time dyslexic, has been using dictation software since the sixth grade when his dad set him up on Dragon Dictation. He used it through school to write papers, and has been keeping his own transcription benchmarks since college. All that time, writing with your voice has remained a cumbersome and brittle experience that is riddled with painpoints.
Dictation software is still terrible. All the solutions basically compete on accuracy (i.e. speech recognition), but none of them deal with the fundamentally brittle nature of the text that they generate. They don't try to format text correctly and require you to learn a bunch of specialized commands, which often are not worth it. They're not even close to a voice replacement for a keyboard.
Even post LLM, you are limited to a set of specific commands and the most accurate models don’t have any commands. Outside of these rules, the models have no sense for what is an instruction and what is content. You can’t say “and format this like an email” or “make the last bullet point shorter”. Aqua solves this.
This problem is important to Finn and millions of other people who would write with their voice if they could. Initially, we didn't think of it as a startup project. It was just something we wanted for ourselves. We thought maybe we'd write a novel with it - or something. After friends started asking to use the early versions of Aqua, it occurred to us that, if we didn't build it, maybe nobody would.
Aqua Voice is a text editor that you talk to like a person. Depending on the way that you say it and the context in which you're operating, Aqua decides whether to transcribe what you said verbatim, execute a command, or subtly modify what you said into what you meant to write.
For example, if you were to dictate: "Gryphons have classic forms resembling shield volcanoes," Aqua would output your text verbatim. But if you stumble over your words or start a sentence over a few times, Aqua is smart enough to figure that out and to only take the last version of the sentence.
The vision is not only to provide a more natural dictation experience, but to enable for the first time an AI-writing experience that feels natural and collaborative. This requires moving away from using LLMs for one-off chat requests and towards something that is more like streaming where you are in constant contact with the model. Voice is the natural medium for this.
Aqua is actually 6 models working together to transcribe, interpret, and rewrite the document according to your intent. Technically, executing a real-time voice application with a language model at its core requires complex coordination between multiple pieces. We use MoE transcription to outperform what was previously thought possible in terms of real-time accuracy. Then we sync up with a language model to determine what should be on the screen as quickly as possible.
The model isn't perfect, but it is ready for early adopters and we’ve already been getting feedback from grateful users. For example, a historian with carpal tunnel sent us an email he wrote using Aqua and said that he is now able to be five times as productive as he was previously. We've heard from other people with disabilities that prevent them from typing. We've also seen good adoption from people who are dyslexic or simply prefer talking to typing. It’s being used for everything from emails to brainstorming to papers to legal briefings.
While there is much left to do in terms of latency and robustness, the best experiences with Aqua are beginning to feel magical. We would love for you to try it out and give us feedback, which you can do with no account on https://withaqua.com. If you find it useful, it’s $10/month after a 1000-token free trial. (We want to bump the free trial in the future, but we're a small team, and running this thing isn’t cheap.)
We’d love to hear your ideas and comments with voice-to-text!