AI was once again the buzzword for this final group of YC Demo Day companies. Nearly 190 companies participated in Y Combinator’s Winter ’26 cohort and showcased their startups during a demo day on Tuesday.
These companies work on products in industries such as law, transportation, and healthcare.
Admittedly, I didn’t listen to every product presentation, given the size of the cohort and this year’s Demo Day format offered to the media. YC released the presentation videos, one by one, approximately 20 minutes after the founders’ presentation (rather than a live stream or invitation to an in-person session).
Instead, I read the pitches of all 190 startups and spent the day watching the pitches of those I found intriguing, then narrowed them down to the 16 that stood out as the most interesting startups in this overflowing YC class.
ARC Prize Foundation
What it does: Creates benchmarks to help measure progress toward AGI.
Why it’s interesting: A non-profit organization in YC! But then again, given that OpenAI, Anthropic, and Google already use some form of the organization’s benchmark, it makes sense to include them. This foundation aims to inspire more open source AGI research by hosting competitions and awarding research grants. One of the reasons for this AI revolution is to achieve AGI (which Nvidia CEO Jensen Huang says is already here), and it will be a matter of historical record to track how close we are to AI machines with general intelligence.
Asimov
What it does: Collecting human movement data to form humanoids
Techcrunch event
San Francisco, California | October 13-15, 2026
Why it’s interesting: People from all over the world submit videos of themselves performing movements and tasks to this company so that it can then turn them into data sets that can help train robots. It’s part of the movement that’s trying to make humanoids a thing, finding uses for them beyond the supply chain and entertainment. I’m optimistic about humanoid technology, even though our “Rosey the Robot” era may still be far away. Using data to teach humanoids the flow and – dare I say, elegance – of human movement could help them be less robotic in performing their tasks.
Voice
What it does: Helps automate tedious non-design work for architectural firms
Why it’s interesting: It’s not every day that we hear about new technologies targeting the architecture sector. The founders themselves observed that this market is underserved (although full of potential). This tool uses AI to help automate tasks that creative types like architects may find tedious, like reviewing specifications, drawings, contracts and proposals.
Computer Button
What it does: Wearable AI
Why it’s interesting: Everyone is trying to make wearable AI a thing as the world waits for OpenAI’s product from its acquisition of Johnny Ive’s company. Two former Applers (can I call them that?) came together to launch Button, essentially a small computer, the founders explained, designed for AI. Button connects to apps like Email, Slack, and Salesforce and operates them via voice command to perform certain tasks. The next must-have hardware will likely be some form of wearable AI, so it’s interesting to see what emerges.
CodeWisp
What it does: Allows anyone to create games using AI
Why it’s interesting: The founders say all you have to do is tell an AI how to make a game, and it will make the game. It’s fun, creative, exciting! I tried making games when I was a kid and always found it difficult and tedious, but the excitement of designing one never went away. As flavor coding has become all the rage for building apps, tools like this make imaginative execution much easier. Perhaps this will be the next generation of atmosphere creation.
Multilayer laboratories
What it does: Help detect website spoofing
Why it’s interesting: The rise of agent tools means that websites are increasingly easy to spoof, and bad guys are quick to use this technology to scam people. Crosslayer Labs helps its customers detect and monitor their online setups so they can stay protected against this nuance of emerging Internet threats.
Doomersion
What it does: Teaches you languages while you doomscroll
Why it’s interesting: We spend far too much time doomscrolling, simply filling our brains with garbage that ends up irritating us or atrophying our brains. This startup is an app that shows users short videos, like how they would view and scroll through them on a TikTok feed, in the language they are trying to learn. If there were to be meaning to all the content we put into our heads all day, what a brilliant way to find it. It combines what consumers won’t stop doing (spending hours on their phones browsing content) with something as cool as learning a language. Very interesting.
Lexius
What it does: Integrates advanced AI into security systems
Why it’s interesting: This uses AI to enhance existing security camera systems, enabling footage to detect and report instances of theft or falls, replacing a fragmented and mostly manual process. The startup says it is targeting businesses that have cameras without artificial intelligence, where a camera could detect an incident, but the company is slow to take action.
Balance Laboratories
What it does: An AI-powered library management system tool
Why it’s interesting: This is AI hitting one of those industries that the tech industry often overlooks: libraries. This startup has created an AI-powered library management system to specifically help schools with inventory and cataloging, right now. As the founder said in his pitch, there isn’t much competition when it comes to automating or innovating the tools already used in this area, which makes any new idea a contender for “the next big thing.”
Milliray
What it does: A radar system to help track small drones
Why it’s interesting: Defense technology is currently one of the hottest technology categories. As the founder of this company said, right now people are out in the fields doing everything they can to track tiny drones, but the human eye can miss a lot or mistake a small drone for a bird and vice versa. This startup uses sensors to identify what is actually a small drone in the sky. Given the current geopolitical state, new technologies are necessary – and emerging – for every country to stay one step ahead of its enemies.
MouseCat
What it does: Uses AI to investigate fraud
Why it’s interesting: AI is a great work tool, and this is true for those whose job it is to defraud and swindle. This company pulls a company’s data from a large cloud storage like Databricks or Snowflake, analyzes the data and consumer activity for anything suspicious, and gives recommendations on how to act. Native AI tools like this are important when it comes to tracking bad AI, which is also capable of unleashing the evil that it is.
Opalite Health
What it does: Uses AI to help healthcare providers speak to non-English speakers
Why it’s interesting: There is a lot of room for interpretation when two people cannot understand each other. In the medical world, it can be a matter of life and death. This AI medical translator helps break the language barrier, allowing healthcare providers to understand patients who speak a different language. In a globalized world (and in a country as diverse as America), it’s important that people can access the health care they need, regardless of language. Of course, this idea isn’t entirely original, as a number of other health tech startups and providers offer a similar service.
Sequential markets
What it does: Let’s trade in different markets, like crypto and prediction, on one system.
Why it’s interesting: As someone who likes everything in one place, I understand the desire for a less fragmented process when executing trades in these markets. It’s the same reason I always like to go to big box retailers to buy brands (I like to see all my options at once).
ShoFo
What it does: A video library of practically everything
Why it’s interesting: This startup bills itself as the “world’s video library,” which is pretty cool in my opinion. I grew up on YouTube and Tumblr and I remember how tricky search was back then, when you were looking for something specific. While this is more of a custom video index to help AI Labs efficiently find various datasets, I like any tool that makes it easier to find and organize.
Sonar
What it does: Helps the software solve its own production problems
Why it’s interesting: This startup is creating technology that seems very cool. It connects to other monitoring systems, promises to reduce alert noise (a distraction from finding the really important alerts), automatically identifies the root causes of problems, then finds ways to resolve them or suggest other actions to engineers. Although there are a growing number of AI code review startups (and this functionality is also offered by modelers), there should be room for independent startups once the code reaches production systems. This is yet another aspect of workflow that founders are automating.
AI Terranox
What it does: Use AI to find uranium deposits in North America
Why it’s interesting: Uranium will be needed to power the next generation of nuclear energy, the company’s founders reminded us. (Nuclear energy is generally considered safe these daysalthough uranium itself is obviously toxic – and the founder did not indicate in his pitch how it could be mined safely.) These founders believe that nuclear power will be necessary to power to us new data centers under construction. Earth will need a lot of energy, from many sources, to ensure its inhabitants can meet the ambitions of the AI revolution.