The best Hacker News stories from Show from the past day

from

All Show

from the past*

day week month

with these search terms:

Create follow Go back

Latest posts:

Show HN: Verify LLM Generated Code with a Spreadsheet

Hey HN! Been a minute. We launched Mito here last year (<a href="https://news.ycombinator.com/item?id=32723766" rel="nofollow">https://news.ycombinator.com/item?id=32723766</a>).<p>Mito is a spreadsheet that generates Python code as you edit it. We've spent the past three years trying to lower the startup cost to use Python for data work. In doing so, we’ve been thrust into the middle of many Python transition processes at larger enterprises, and we’ve seen up-close how non-technical folks interact with generated code.<p>The Mito AI chatbot lives inside of the Mito spreadsheet (<a href="https://www.trymito.io/">https://www.trymito.io/</a>>. The obvious benefit of this is that you can use the chatbot to transform your data and write a repeatable Pythons script. The less obvious (but equally important) benefit is that by connecting a spreadsheet and chatbot, Mito helps you understand the impact of your edits and verify LLM generated code. Every time you use the chatbot, Mito highlights the changed data in the spreadsheet. You can see a quick demo here (<a href="https://www.tella.tv/video/clibtwssv00000fl65oky13nu/view">https://www.tella.tv/video/clibtwssv00000fl65oky13nu/view</a>).<p>Three main insights shaped our approach to LLM code generation:<p># Consumers of generated code don't know enough Python to verify and correct the code<p>Mito users span the range of Python experience. For new programmers, generating code using LLMs is an easy step one. Ensuring the generated code is correct is the forgotten step two.<p>In practice, LLMs often generate incorrect code, or code with unexpected side effects. A user will prompt an LLM to calculate a total_revenue column from price and quantity columns. The LLM correctly calculates total_revenue = price * quantity but then mistakenly deletes price and quantity.<p>New programmers find it almost impossible to verify generated code by reading it alone. They need tooling designed for their skillsets.<p># Not everyone knows how to use a chat interface for transformations<p>We were surprised to learn that many Mito users a) had no experience with ChatGPT, and b) didn’t understand the chat interface at all! Mito AI presents users a few example prompts and an input field. A surprising number of users thought the example prompts were all they could use Mito AI for.<p>AI chatbots are new. Us builders might be using them for natural language interactions, but users are still learning how to use them in new contexts. This stands in stark contrast to spreadsheets, where pretty much ever business user has experience. Shout out 40 years of Excel dominance!<p># The more context a prompt has about the user’s data + edits, the better the LLM results<p>For the LLM to generate code that can execute correctly, the prompt should include the names of the dataframes, the column headers, (some) dataframe values, and a few previous edits as examples. Duh.<p>But there’s no reason users should be responsible for writing this prompt. No one loves writing long chats, and in practice Mito AI users expect to be able to write ~12 words. Spreadsheets are well-suited to building the rest of the prompt for you - they have all of your data context, and know your recent edits.<p>With these three insights, it became very clear to us what role a spreadsheet could play in LLM based code-gen: a spreadsheet is the prompt builder, and a spreadsheet is the code verifier.<p>Mito AI builds an effective prompt by supplementing your input with the context of your data and recent edits.<p>Mito AI then helps you to verify the LLM generated code by highlighting the added, modified, and removed data within the chat interface - and within the spreadsheet. This way, you can ensure your LLM generated code is correct.<p>Give it a spin. Let us know what you think of the recon and how we can make it more helpful!<p>Also, if you like what we’re doing, we’re hiring – come help us build! (<a href="https://www.ycombinator.com/companies/mito/jobs" rel="nofollow">https://www.ycombinator.com/companies/mito/jobs</a>)

Show HN: Verify LLM Generated Code with a Spreadsheet

Hey HN! Been a minute. We launched Mito here last year (<a href="https://news.ycombinator.com/item?id=32723766" rel="nofollow">https://news.ycombinator.com/item?id=32723766</a>).<p>Mito is a spreadsheet that generates Python code as you edit it. We've spent the past three years trying to lower the startup cost to use Python for data work. In doing so, we’ve been thrust into the middle of many Python transition processes at larger enterprises, and we’ve seen up-close how non-technical folks interact with generated code.<p>The Mito AI chatbot lives inside of the Mito spreadsheet (<a href="https://www.trymito.io/">https://www.trymito.io/</a>>. The obvious benefit of this is that you can use the chatbot to transform your data and write a repeatable Pythons script. The less obvious (but equally important) benefit is that by connecting a spreadsheet and chatbot, Mito helps you understand the impact of your edits and verify LLM generated code. Every time you use the chatbot, Mito highlights the changed data in the spreadsheet. You can see a quick demo here (<a href="https://www.tella.tv/video/clibtwssv00000fl65oky13nu/view">https://www.tella.tv/video/clibtwssv00000fl65oky13nu/view</a>).<p>Three main insights shaped our approach to LLM code generation:<p># Consumers of generated code don't know enough Python to verify and correct the code<p>Mito users span the range of Python experience. For new programmers, generating code using LLMs is an easy step one. Ensuring the generated code is correct is the forgotten step two.<p>In practice, LLMs often generate incorrect code, or code with unexpected side effects. A user will prompt an LLM to calculate a total_revenue column from price and quantity columns. The LLM correctly calculates total_revenue = price * quantity but then mistakenly deletes price and quantity.<p>New programmers find it almost impossible to verify generated code by reading it alone. They need tooling designed for their skillsets.<p># Not everyone knows how to use a chat interface for transformations<p>We were surprised to learn that many Mito users a) had no experience with ChatGPT, and b) didn’t understand the chat interface at all! Mito AI presents users a few example prompts and an input field. A surprising number of users thought the example prompts were all they could use Mito AI for.<p>AI chatbots are new. Us builders might be using them for natural language interactions, but users are still learning how to use them in new contexts. This stands in stark contrast to spreadsheets, where pretty much ever business user has experience. Shout out 40 years of Excel dominance!<p># The more context a prompt has about the user’s data + edits, the better the LLM results<p>For the LLM to generate code that can execute correctly, the prompt should include the names of the dataframes, the column headers, (some) dataframe values, and a few previous edits as examples. Duh.<p>But there’s no reason users should be responsible for writing this prompt. No one loves writing long chats, and in practice Mito AI users expect to be able to write ~12 words. Spreadsheets are well-suited to building the rest of the prompt for you - they have all of your data context, and know your recent edits.<p>With these three insights, it became very clear to us what role a spreadsheet could play in LLM based code-gen: a spreadsheet is the prompt builder, and a spreadsheet is the code verifier.<p>Mito AI builds an effective prompt by supplementing your input with the context of your data and recent edits.<p>Mito AI then helps you to verify the LLM generated code by highlighting the added, modified, and removed data within the chat interface - and within the spreadsheet. This way, you can ensure your LLM generated code is correct.<p>Give it a spin. Let us know what you think of the recon and how we can make it more helpful!<p>Also, if you like what we’re doing, we’re hiring – come help us build! (<a href="https://www.ycombinator.com/companies/mito/jobs" rel="nofollow">https://www.ycombinator.com/companies/mito/jobs</a>)

Show HN: Verify LLM Generated Code with a Spreadsheet

Hey HN! Been a minute. We launched Mito here last year (<a href="https://news.ycombinator.com/item?id=32723766" rel="nofollow">https://news.ycombinator.com/item?id=32723766</a>).<p>Mito is a spreadsheet that generates Python code as you edit it. We've spent the past three years trying to lower the startup cost to use Python for data work. In doing so, we’ve been thrust into the middle of many Python transition processes at larger enterprises, and we’ve seen up-close how non-technical folks interact with generated code.<p>The Mito AI chatbot lives inside of the Mito spreadsheet (<a href="https://www.trymito.io/">https://www.trymito.io/</a>>. The obvious benefit of this is that you can use the chatbot to transform your data and write a repeatable Pythons script. The less obvious (but equally important) benefit is that by connecting a spreadsheet and chatbot, Mito helps you understand the impact of your edits and verify LLM generated code. Every time you use the chatbot, Mito highlights the changed data in the spreadsheet. You can see a quick demo here (<a href="https://www.tella.tv/video/clibtwssv00000fl65oky13nu/view">https://www.tella.tv/video/clibtwssv00000fl65oky13nu/view</a>).<p>Three main insights shaped our approach to LLM code generation:<p># Consumers of generated code don't know enough Python to verify and correct the code<p>Mito users span the range of Python experience. For new programmers, generating code using LLMs is an easy step one. Ensuring the generated code is correct is the forgotten step two.<p>In practice, LLMs often generate incorrect code, or code with unexpected side effects. A user will prompt an LLM to calculate a total_revenue column from price and quantity columns. The LLM correctly calculates total_revenue = price * quantity but then mistakenly deletes price and quantity.<p>New programmers find it almost impossible to verify generated code by reading it alone. They need tooling designed for their skillsets.<p># Not everyone knows how to use a chat interface for transformations<p>We were surprised to learn that many Mito users a) had no experience with ChatGPT, and b) didn’t understand the chat interface at all! Mito AI presents users a few example prompts and an input field. A surprising number of users thought the example prompts were all they could use Mito AI for.<p>AI chatbots are new. Us builders might be using them for natural language interactions, but users are still learning how to use them in new contexts. This stands in stark contrast to spreadsheets, where pretty much ever business user has experience. Shout out 40 years of Excel dominance!<p># The more context a prompt has about the user’s data + edits, the better the LLM results<p>For the LLM to generate code that can execute correctly, the prompt should include the names of the dataframes, the column headers, (some) dataframe values, and a few previous edits as examples. Duh.<p>But there’s no reason users should be responsible for writing this prompt. No one loves writing long chats, and in practice Mito AI users expect to be able to write ~12 words. Spreadsheets are well-suited to building the rest of the prompt for you - they have all of your data context, and know your recent edits.<p>With these three insights, it became very clear to us what role a spreadsheet could play in LLM based code-gen: a spreadsheet is the prompt builder, and a spreadsheet is the code verifier.<p>Mito AI builds an effective prompt by supplementing your input with the context of your data and recent edits.<p>Mito AI then helps you to verify the LLM generated code by highlighting the added, modified, and removed data within the chat interface - and within the spreadsheet. This way, you can ensure your LLM generated code is correct.<p>Give it a spin. Let us know what you think of the recon and how we can make it more helpful!<p>Also, if you like what we’re doing, we’re hiring – come help us build! (<a href="https://www.ycombinator.com/companies/mito/jobs" rel="nofollow">https://www.ycombinator.com/companies/mito/jobs</a>)

Show HN: Automating daily reports, because fuck it

Show HN: Automating daily reports, because fuck it

Show HN: Automating daily reports, because fuck it

Show HN: Mercury – Convert Jupyter notebooks to web apps

Author here. Mercury is the simplest way to serve your notebooks as web apps. The simplicity of the framework is very important to us. Mercury has some useful features to make sharing easier:<p>- you can show or hide your code,<p>- your users can easily export executed notebook to PDF/HTML,<p>- there is built-in authentication,<p>- you can produce files in the notebook and make them downloadable,<p>- you can share multiple notebooks.<p>We also care about deployment simplicity. That's why we created a shared hosting service called Mercury Cloud. You can deploy notebook by uploading a file. Below clickable links:<p>The GitHub repository <a href="https://github.com/mljar/mercury">https://github.com/mljar/mercury</a><p>Documentation <a href="https://RunMercury.com/docs/" rel="nofollow">https://RunMercury.com/docs/</a><p>Mercury Cloud <a href="https://cloud.runmercury.com" rel="nofollow">https://cloud.runmercury.com</a>

Show HN: Mercury – Convert Jupyter notebooks to web apps

Author here. Mercury is the simplest way to serve your notebooks as web apps. The simplicity of the framework is very important to us. Mercury has some useful features to make sharing easier:<p>- you can show or hide your code,<p>- your users can easily export executed notebook to PDF/HTML,<p>- there is built-in authentication,<p>- you can produce files in the notebook and make them downloadable,<p>- you can share multiple notebooks.<p>We also care about deployment simplicity. That's why we created a shared hosting service called Mercury Cloud. You can deploy notebook by uploading a file. Below clickable links:<p>The GitHub repository <a href="https://github.com/mljar/mercury">https://github.com/mljar/mercury</a><p>Documentation <a href="https://RunMercury.com/docs/" rel="nofollow">https://RunMercury.com/docs/</a><p>Mercury Cloud <a href="https://cloud.runmercury.com" rel="nofollow">https://cloud.runmercury.com</a>

Show HN: HN Resume to Jobs – AI Powered Job Matching Tailored to Your Resume

Hey HN! I'm excited to show off this side project I've been working on. This project matches your resume with the best matching jobs from the monthly HN Who's Hiring post. It works by creating a vector embedding of your resume using OpenAI's embedding API, and then ranking the jobs using a vector similarity score. (You can toggle between max inner product, cosine, and euclidean in the "Advanced Options")<p>I was laid off in August and it took a whole 6 months for me to find my new job. Fortunately, I found my new role on January's HN Who's Hiring post. So I hope this will prove useful to any job seekers out there. I know it's a tough time right now, but you will get through it!<p>Thanks HN! I would greatly appreciate any and all feedback!

Show HN: HN Resume to Jobs – AI Powered Job Matching Tailored to Your Resume

Hey HN! I'm excited to show off this side project I've been working on. This project matches your resume with the best matching jobs from the monthly HN Who's Hiring post. It works by creating a vector embedding of your resume using OpenAI's embedding API, and then ranking the jobs using a vector similarity score. (You can toggle between max inner product, cosine, and euclidean in the "Advanced Options")<p>I was laid off in August and it took a whole 6 months for me to find my new job. Fortunately, I found my new role on January's HN Who's Hiring post. So I hope this will prove useful to any job seekers out there. I know it's a tough time right now, but you will get through it!<p>Thanks HN! I would greatly appreciate any and all feedback!

Show HN: HN Resume to Jobs – AI Powered Job Matching Tailored to Your Resume

Hey HN! I'm excited to show off this side project I've been working on. This project matches your resume with the best matching jobs from the monthly HN Who's Hiring post. It works by creating a vector embedding of your resume using OpenAI's embedding API, and then ranking the jobs using a vector similarity score. (You can toggle between max inner product, cosine, and euclidean in the "Advanced Options")<p>I was laid off in August and it took a whole 6 months for me to find my new job. Fortunately, I found my new role on January's HN Who's Hiring post. So I hope this will prove useful to any job seekers out there. I know it's a tough time right now, but you will get through it!<p>Thanks HN! I would greatly appreciate any and all feedback!

Show HN: Minutes – Save up to 20% of salespeople's time

Hey everyone, I wanted to share something I've been working on that I think could be really helpful for sales teams. Minutes is a startup that utilizes OpenAI technology to automate the transcription and summarization of sales calls.<p>What Minutes is doing can be described in 3 steps: 1. Taking your audio or video and transcribing it to text 2. Taking the most important and relevant information and creating a summary, including key points and the next tasks 3. Creating a follow-up email, including the summary points and next steps<p>Story behind it: As an entrepreneur who handles sales and marketing, I noticed a common challenge: the amount of time managers spend on creating call summaries and writing follow-up emails. Not only is it time-consuming, but it's also often a hassle to ensure accuracy. To tackle this issue, I decided to delve deeper and spoke with several sales directors. Turns out, this problem is widespread, and even salespeople themselves dislike the administrative burden it brings. Inspired by these insights, I created Minutes.<p>Call to action: Here's where you come in. Sign up for Minutes using the code "minutes_alpha," and you'll receive 60 free transcription minutes to play with the service. I genuinely value your feedback and insights, as they will help us refine Minutes and tailor it to the specific needs of sales teams everywhere.<p>Thank you for your support, and I'm eagerly looking forward to hearing your thoughts.<p>Warm regards, Bogdan, founder of Minutes

Show HN: Minutes – Save up to 20% of salespeople's time

Hey everyone, I wanted to share something I've been working on that I think could be really helpful for sales teams. Minutes is a startup that utilizes OpenAI technology to automate the transcription and summarization of sales calls.<p>What Minutes is doing can be described in 3 steps: 1. Taking your audio or video and transcribing it to text 2. Taking the most important and relevant information and creating a summary, including key points and the next tasks 3. Creating a follow-up email, including the summary points and next steps<p>Story behind it: As an entrepreneur who handles sales and marketing, I noticed a common challenge: the amount of time managers spend on creating call summaries and writing follow-up emails. Not only is it time-consuming, but it's also often a hassle to ensure accuracy. To tackle this issue, I decided to delve deeper and spoke with several sales directors. Turns out, this problem is widespread, and even salespeople themselves dislike the administrative burden it brings. Inspired by these insights, I created Minutes.<p>Call to action: Here's where you come in. Sign up for Minutes using the code "minutes_alpha," and you'll receive 60 free transcription minutes to play with the service. I genuinely value your feedback and insights, as they will help us refine Minutes and tailor it to the specific needs of sales teams everywhere.<p>Thank you for your support, and I'm eagerly looking forward to hearing your thoughts.<p>Warm regards, Bogdan, founder of Minutes

Show HN: Git credential helper using OAuth in browser

I authenticate to many Git hosts from many machines and got tired of generating and copying personal access tokens. With credential helper git-credential-oauth, there are no personal access tokens or SSH keys to configure. Instead you authenticate in browser using OAuth.<p>Git Credential Manager (included with Git for Windows) has a similar feature but it's awkward for Linux users to install. git-credential-oauth is cross platform and packaged in many Linux distributions.

Show HN: Git credential helper using OAuth in browser

I authenticate to many Git hosts from many machines and got tired of generating and copying personal access tokens. With credential helper git-credential-oauth, there are no personal access tokens or SSH keys to configure. Instead you authenticate in browser using OAuth.<p>Git Credential Manager (included with Git for Windows) has a similar feature but it's awkward for Linux users to install. git-credential-oauth is cross platform and packaged in many Linux distributions.

Show HN: Word2vec Algorithm in ~100sloc with NumPy

Here's a small demonstration of the fundamental aspects of the word-to-vec algorithm. It's implemented in a single python script and depends only on a single text file for training.<p>It's not meant to be blazingly fast or anything, just a toy example to aid my understanding of how word vectors might be learnt from a corpus.

Show HN: Word2vec Algorithm in ~100sloc with NumPy

Here's a small demonstration of the fundamental aspects of the word-to-vec algorithm. It's implemented in a single python script and depends only on a single text file for training.<p>It's not meant to be blazingly fast or anything, just a toy example to aid my understanding of how word vectors might be learnt from a corpus.

Show HN: Discipline.io – Make binding commitments to your better self

Hi HN,<p>It's super hard to quit addicting apps. Existing solutions are non-binding, and therefore, require constant self-discipline and vigilance to be effective. Which is ironic, considering the problem we are trying to solve, is a lack of self-discipline.<p>There exists a physical solution to this problem, in the form of a time-locked safe[1]. It’s effective for reducing app usage, but it's cumbersome and inconvenient.<p>Fundamentally, what these lock-boxes do is impose a real-world cost on breaking your commitment. It’s not impossible to break your commitment, but it would require destroying the lock-box, and that incurs a cost.<p>I created a digital version of this, as an app (Android-only). Commitment’s are backed with cash deposits, and the phone API’s are used to detect violations and enforce compliance. No self-reporting required. And it even works if the monitoring app is uninstalled, or its background service disabled.<p>An example commitment: Stop using Tinder, for the next week, or forfeit $20 to the Red Cross charity.<p>The solution also generalizes to other types of commitments:<p><pre><code> - Commit to using an app more (e.g. meditation) - Commit to visiting a location (e.g. gym) - Commit to making a phone call (e.g. mom) </code></pre> It’s currently only available for Android, but I'm considering doing an IOS version as well. Hope you like it, and please share any feedback you might have in the comments!<p>[1]: <a href="https://www.thekitchensafe.com" rel="nofollow">https://www.thekitchensafe.com</a>

Show HN: Discipline.io – Make binding commitments to your better self

Hi HN,<p>It's super hard to quit addicting apps. Existing solutions are non-binding, and therefore, require constant self-discipline and vigilance to be effective. Which is ironic, considering the problem we are trying to solve, is a lack of self-discipline.<p>There exists a physical solution to this problem, in the form of a time-locked safe[1]. It’s effective for reducing app usage, but it's cumbersome and inconvenient.<p>Fundamentally, what these lock-boxes do is impose a real-world cost on breaking your commitment. It’s not impossible to break your commitment, but it would require destroying the lock-box, and that incurs a cost.<p>I created a digital version of this, as an app (Android-only). Commitment’s are backed with cash deposits, and the phone API’s are used to detect violations and enforce compliance. No self-reporting required. And it even works if the monitoring app is uninstalled, or its background service disabled.<p>An example commitment: Stop using Tinder, for the next week, or forfeit $20 to the Red Cross charity.<p>The solution also generalizes to other types of commitments:<p><pre><code> - Commit to using an app more (e.g. meditation) - Commit to visiting a location (e.g. gym) - Commit to making a phone call (e.g. mom) </code></pre> It’s currently only available for Android, but I'm considering doing an IOS version as well. Hope you like it, and please share any feedback you might have in the comments!<p>[1]: <a href="https://www.thekitchensafe.com" rel="nofollow">https://www.thekitchensafe.com</a>

Show HN: StonksGPT – A Natural Language search tool for Stocks and Finance data