MinusX

MinusX

Software Development

San Francisco, CA 453 followers

Let our AI Data Scientist use your analytics tools (Jupyter, Grafana, Metabase, etc.) for you.

About us

Website
https://minusx.ai
Industry
Software Development
Company size
2-10 employees
Headquarters
San Francisco, CA
Type
Privately Held

Locations

Employees at MinusX

Updates

  • Let the Plots unfold!

    View profile for Vivek Aithal, graphic

    CEO @MinusX (YC S24); Ex-Research Engineer @comma.ai

    Today we're launching Plots - a data story magazine powered by MinusX ! I started doing data viz using D3js ~2014. I'm as fascinated and starry eyed now about data stories as I was then. After running NumbersOfIndia (https://numbersofindia.com) for a brief while, poring hours into each story, I realized how many more beautiful stories would exist if it was easier to create them. Using MinusX has made this process delightful to me. We're going to use Plots as a medium where curiosity meets MinusX - covering topics in sports, politics, science and society. Here's the post announcing this launch: https://lnkd.in/grKUBbTD . Subscribe on there so that you can get the next Plot in your inbox! Also shoutout to Akash Mukherjee, The Pudding, and Mike Bostock for all the inspiration! Let me know what Plot you want to see!

    • No alternative text description for this image
  • MinusX reposted this

    View profile for Sreejith Puthanpurayil, graphic

    Co-Founder & CTO MinusX.ai (YC S24). Ex-Facebook, GoJek

    Level up your Google Sheets with an AI data analyst by your side! Excited to announce that MinusX now adds on to Google Sheets with a simple one-click install: https://lnkd.in/gsCN5dAr We've been working on this for some time and we're super interested in how y'all use it :) We're still actively developing so please send me any feedback/suggestions and we'll try to incorporate it! Here's a glimpse at what we can do right now:

  • We just open-sourced the entire MinusX chrome extension on GitHub under MIT License! The three main motivations for this decision were: Trust, Extensibility and Incentive Alignment. Trust: Given some of the shenanigans we've seen some of these browser automation companies do (all the worst things you can imagine 👀 ), we're honestly shocked anyone uses any extension that is not open source. Extensibility: We imagine almost all software we use today to either be replaced, or retrofit with AI. The retrofitting will not always be done by the makers of the original software. We're inspired by comma.ai and even have adopted their opendbc folder structure! Incentive Alignment: We don't want to lock you in, we don't want to sell your data. We want you to have the best data science assistant sitting with you. And when you win, we win. :) Read more about it in our blogpost: https://lnkd.in/gv4bF5cr GitHub Repo: https://lnkd.in/gnjBsGuk

    • No alternative text description for this image
  • Gain insights fast, with MinusX!

    View profile for Vivek Aithal, graphic

    CEO @MinusX (YC S24); Ex-Research Engineer @comma.ai

    Cost/accuracy of models (by OpenAI, Anthropic) are well reported on, but *latency* is much more important while building applications like minusx.ai. Here's a quick analysis (done completely using MinusX) and the implications. Consider a simple linear regression fit: for gpt-4o each output token is ~250x as slow as one input token and for claude-3-5-sonnet, it is ~115x! How can you use this information in your AI app design? 1. Give ~100x more input tokens if it means you can reduce 1 output token. "No yapping" is good. 2. Chain of Thought is not a free lunch! Adding extra "CoT" tokens adds massive latency. If you can achieve the same accuracy, multi-turn may turn out to be much faster than CoT despite adding an extra API call! 3. Sequential tool call may be better than Parallel. Most of the latency is from the output tokens and the input tokens are almost a rounding error. So even though parallel tool calls seem like an improvement, sequential calls result in a much snappier experience. I did this whole analysis on Jupyter using minusx.ai (in ~8.5 mins). If you use Jupyter (or Metabase) for analytics, give MinusX a try! Check out the blogpost for more analysis, the notebook and data (linked in the comment)

    • No alternative text description for this image
    • No alternative text description for this image
    • No alternative text description for this image
    • No alternative text description for this image
  • When we say we integrate deeply into tools, we mean business. MinusX knows your Metabase inside out, better than anyone! Imagine the collective knowledge of all your colleagues, their experiments, queries and dashboards powering your next exploration. Want to navigate thousands of tables to just get your user retention? Ask MinusX! Want to drill down in any question, hit ⌘+k, and ask away! Want to understand a complicated SQL that has been passed down many generations of analysts (we see you 👀 ), select the region and ask MinusX! If you use Metabase for your everyday tasks, try MinusX out (https://minusx.ai)! You can use it on your own data, in your own Metabase instance in <2 mins!

    • No alternative text description for this image
  • View organization page for MinusX, graphic

    453 followers

    MinusX was born out of our ❤️ for Project Jupyter. Our whole shtick is that tools we use today for data science / data analysis are incredible. With the right user ( 👀 ), you can do pretty much anything you want with them! Want to explore data? Just ask MinusX in a notebook! Want to pair with an AI Data Scientist? Hit ⌘+k! MinusX patiently waits on the side, refining it's context, learning from you. Just ask! Or want to understand a messy notebook/unclear plot someone sent you? Select the region, and - as always - ask MinusX! If you use Jupyter for your everyday tasks, try MinusX out (https://minusx.ai)! You can use it on your own data, in your own Jupyter instance in <2 mins!

    • No alternative text description for this image
  • Over the last couple of weeks, you may have seen us launching MinusX on YC, Hacker News, Product Hunt, and a whole lot of other places. Now that there's a bit of calm (read as dealing with only a handful raging fires), here's a bit more context on the problem we're working on, our take on the solution surface, and why we're excited to be building MinusX. Getting, analyzing, and communicating accurate data is a struggle in orgs of all sizes and shapes. AI native solutions that require massive upfront costs and re-learning workflows are a non starter. We think smart AI agents should just work in tools you already use, and love. MinusX is exactly this. It just works. Isn't this what a smart colleague would do? Read more below, and try us out! https://lnkd.in/gdFcCGxF

    Welcome to MinusX!

    Welcome to MinusX!

    minusx.ai

  • We're super thrilled to launch on YC! We're going to build the best data science agent out there! You can try MinusX today on Jupyter and Metabase! Head on over to https://minusx.ai for more info on how to get started!

    View organization page for Y Combinator, graphic

    1,117,212 followers

    minusx.ai (YC S24) is a Chrome extension that adds a side chat to Jupyter and Metabase. Given an instruction, their agent operates your apps to analyze data and answer queries. Their AI data scientist is a scientist, not yet-another-new-analytics-platform. MinusX just works, in tools you already use and love. You can: 1. Generate hypotheses and explore data 2. Interop with MinusX to modify existing Jupyter notebooks or Metabase questions 3. Select a region and ask questions, or ask for modifications The lack of data analyst/data scientist bandwidth is a real pain. If you’re a programmer, you just want answers. If you’re a product manager, you just want answers. If you’re an analyst/scientist, you want 10 clones of yourself. Any new fancy “talk-to-data” platform requires you to migrate your data and convince your whole team to move over— just to start using it. MinusX solves all these and more, and generally gets out of the way. The founders, Vivek Aithal, Sreejith Puthanpurayil, and Arpit Saxena, have extensive experience in machine learning, data infrastructure, and systems engineering from their previous roles at companies like comma.ai, Meta, and Udaan. They strongly believe that you already have all the tools you need; you just need something like MinusX to use them. Congrats to the team on the launch! 🚀 https://lnkd.in/gNVYJijt

Similar pages

Browse jobs

Funding

MinusX 1 total round

Last Round

Pre seed

US$ 500.0K

Investors

Y Combinator
See more info on crunchbase