TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Show HN: Snapsked – Turn screenshots & photos into action items on calendar

1 点作者 notnotrishi大约 1 年前
Like many, I save a ton of screenshots and photos to remember things -- and then forget lol.<p>Snapsked is a simple tool that takes images, extracts action items, and syncs to your calendar automatically.<p>Some real-life examples: <a href="https:&#x2F;&#x2F;bit.ly&#x2F;snapskeddemo" rel="nofollow">https:&#x2F;&#x2F;bit.ly&#x2F;snapskeddemo</a><p>It’s a small step towards an ambitious idea I’ve had for a while (see background in the comment below if interested): What if an AI could look what you’re looking at (your screens, handwritten notes etc.) while you’re working; understood relevant action items; and helped with things on demand? Just like having a peer look over your shoulder to help when needed.<p>I know it feels like it should be easy to build something like this with vision models available today, but there is one key issue: Models like GPT-4V, even when they work reliably, get expensive soon and are not ideal for end-user use-cases. I spent the last few months working on addressing this key technical problem, and I’ve been able to build an inference flow that works reliably at ~1&#x2F;10th of the cost of GPT-4V, for this specific use case; thus making it affordable for many today.<p>I&#x27;m looking for early users and useful feedback. Sign up at snapsked.org and follow x.com&#x2F;snapsked for updates.<p>Also open for any questions here or on my personal X&#x2F;Twitter (same id as my HN).

2 条评论

notnotrishi大约 1 年前
Also a bit of a background --<p>2018-19: I started working on the problem of automatically extracting all action items from all tools I use (email, slack, GitHub etc.), and getting them in one place. I’d built a few different solutions using NLP (using techniques available back then). Things sort-of worked, but there were always so many edge cases — early users churned soon.<p>2020-21: Graduate student at an AI lab — Mila, Quebec (dropped out eventually) — went there primarily to learn new techniques to build some thing in this space. Around mid-2021 I decided to put this on hold, and went on to work at Beeper (beeper.com) till recently.
mdrzn大约 1 年前
The sign up page is a Google Form? And you ask &quot;What are you comfortable paying per month&quot;? Is there even a real product or is this just a landing page to collect data?
评论 #40155935 未加载