Y Combinator’s Demo Day was a bit different this time around.
As concerns grew over the spread of COVID-19, Y Combinator shifted the event format away from the two-day gathering in San Francisco we’ve gotten used to, instead opting to have its entire class debut to invited investors and media via YC’s Demo Day website.
In a bit of a surprise twist, YC also moved Demo Day forward one week, citing accelerated pacing from investors. Alas, this meant switching up its plan for each company to have a recorded pitch on the Demo Day website; instead, each company pitched via slides, a few paragraphs outlining what they’re doing and the traction they’re seeing, and team bios. It’s unclear so far how this new format — in combination with the rapidly evolving investment climate — will impact this class.
As we do with each class, we’ve collected our notes on each company based on information gathered from their pitches, websites and, in some cases, our earlier coverage of them.
To make things a bit easier to read, we’ve split things up by category rather than have it be one huge wall of text. These are the companies that are working on hardware, robotics, AI, machine learning or tools for developers. You can find the other categories (such as biotech, consumer, and fintech) here.
AI and Machine Learning
Datasaur: A tool meant to help humans label machine data data sets more accurately and efficiently through things like auto-correct, auto-suggest and keyboard hotkeys. It’s free for individual labelers, $100 per month for teams of up to 20 labelers, with custom pricing for larger teams.
1build: Automatic, data-driven job cost estimates for construction companies. You upload your plans, and 1build says it can prepare accurate bids “in minutes.” The company projects a revenue run rate of over $600,000, and says it has completed estimates for mega companies like Amazon, Starbucks and 7-Eleven.
Handl: An API for turning paper documents — including handwritten ones — into structured data ready to be plunked into a database or CRM. While the company says that around 85% of its processing is handled by their AI, it’s backed by humans to validate data when the AI’s confidence is low. Nine months after launch, the company is seeing an ARR of $0.9 million.
Zumo Labs: Uses game engines to generate pre-labeled training data for computer vision systems. By synthesizing the data rather than collecting it from photos/videos of the real world, the company says it can create massive data sets faster, cheaper and without privacy issues.
Teleo: Retrofits existing construction equipment to allow operators to control them remotely. The company says it has built a “fully functional teleoperated loader” since being founded three months ago, and plans to charge construction companies a flat monthly fee per vehicle. The company’s co-founders were previously head of Hardware Engineering and director of Product Manager at Lyft, with both having worked on Google’s Street View team.
Menten AI: Menten AI says it’s using “quantum computing and machine learning” combined with synthetic biology to design new protein-based drugs.
Turing Labs Inc.: Automated, simulated testing of different formulas for consumer goods like soaps and deodorant. Home products and cosmetics can be months of work for R&D labs. Turing has built an AI engine that helps with this process — much like the AI engines used in drug discovery — cutting down the time to days. It’s already working with some of the biggest CPG companies in the world. You can find our previous coverage on Turing here.
Segmed: Segmed is building data sets for AI-driven medical research. Rather than requiring each and every researcher to individually partner with hospitals and imaging facilities, Segmed partners with these organizations (currently over 50) and standardizes, labels and anonymizes the data.
Ardis AI: Ardis AI wants to build the foundation of artificial general intelligence — technology that read and comprehend text like a human. By combining neural networks, symbolic reasoning and new natural language processing techniques, Ardis AI can serve companies that don’t want to hire teams to do data extraction and labeling.
Agnoris: Agnoris analyzes a restaurant’s point-of-sale data to recommend changes to pricing, delivery menus and staffing. For $3,600 per year per restaurant location, Agnoris claims to be able to raise profits by 20%. The company started after the founder opened a restaurant that was packed yet losing money, so it built machine learning tools to improve margins and now it’s selling that software to all eateries.
Froglabs: Froglabs provides weather forecasting AI to businesses for predicting solar and wind energy production, delivery delays, staffing shortages, sales demand and food availability. By ingesting petabytes of weather data, it can save companies money by ensuring their logistics aren’t disrupted. Founded by a long-time Googler who started its Project Loon internet-beaming weather balloons, it’s now signing up e-commerce, retail, rideshare, restaurant and event businesses.
PillarPlus: PillarPlus is a platform that automates the blueprint-designing phase of a building project. It takes a design from an architect or contractor and maps out mechanical, fire, electrical and plumbing details, and estimates the bill of materials and project cost, steps that otherwise take months of work.
Glisten: Glisten uses computer vision and machine learning technologies to develop better, more consistent data sets for e-commerce companies. Its first product is an AI-based tool to populate and enrich sparse product data. Find our previous coverage of Glisten here.
nextmv: Nextmv gives its customers the ability to create their own logistics algorithms automatically — allowing businesses to optimize fleets and manage routes internally.
Visual One: Movement-detecting security cameras can bring up a lot of false positives: there’s motion, yes, but not necessarily anything harmful. Visual One has built an AI platform that integrates with home security cameras to “read” the specific movements that they detect. Owners can create customised alerts so they get notifications only for what they care about. The company’s software can check for furniture-destroying pets, package-lifting thieves, the death-defying antics of toddlers and more. Find our previous coverage of Visual One here.
PostEra: “Medicinal chemistry-as-a-service” is the idea here: PostEra’s platform can design and synthesize molecules faster and at a lower cost than the typical R&D lab, speeding up the research time it takes to test new combinations in the drug discovery process.
Hardware and Robotics
Cyberdontics: Robotics have already revolutionized surgery, courtesy of companies like da Vinci-maker, Intuitive. Cyberdontics is aimed at doing the same for oral surgery, beginning with crowns — one of the more expensive and time-intensive procedures. The company says its robot is capable of performing the generally two-hour procedure in 15 minutes, charging a mere $140 for the job.
Avion: Focused on inhabitants of difficult to reach areas in Africa, Avion is building a drone-based delivery system. The plans consist of medium and long-range medical drones tied to a centralized hub. The drones are hybrid and autonomous with vertical take-off capabilities, able to take 5-kg payloads as far as 150 kms.
SOMATIC: Industrial bathroom cleaning is a prime “dull”/“dirty” candidate to be replaced by automation. Somatic builds large robots that are trained to clean restrooms via VR. The system sprays and wipes down surfaces and is capable of opening doors and riding up and down in the elevator. Find our previous coverage of SOMATIC here.
RoboTire: Anyone who’s ever sat in a service shop waiting room knows how time-intensive the process can be. RoboTire promises to cut the wait time from 60 minutes down to 10 for a set of four tires. The company has begun piloting the technology in locations around the U.S. Find our previous coverage of RoboTire here.
Morphle: Designed to replace outdated analog microscopes, Morphle’s system uses robotic automation to improve imaging. The startup processes higher-resolution images than far pricier systems and with a much smaller failure rate. Morphle has begun selling its system to labs in India.
Daedalus: Founded by an early engineer at OpenAI, Daedalus is building autonomous software to allow industrial robots to operate without human programming, beginning with CNC machines. The company projects that it can improve productivity in the metal machining market by 5x.
Exosonic, Inc.: Exosonic makes supersonic commercial aircraft that don’t have to produce a loud sonic boom, so they can be flown over land. Its goal is a plane that can fly from SF to NYC in three hours. The CEO worked on NASA’s low-boom X-59 aircraft while at Lockheed Martin. Exosonic now has letters of intent from a major airline and two Department of Defense groups, plus a $300,000 U.S. Air Force contract.
Nimbus: Founded by a serial entrepreneur and based in Ann Arbor, Mich., Nimbus is developing the next-generation vehicle platform for urban transportation. Founder Lihang Nong previously launched the fuel-injection systems developer PicoSpray and is now looking to answer the question, “Can a vehicle be several times more space and energy efficient than today’s cars while actually being more comfortable to ride in?”
UrbanKisaan: UrbanKisaan is a vertical farming operation based in India that delivers fresh produce subscriptions to households. Its farms of stacked-up hydroponic tables can be located near cities with just 1% of the land usage of traditional agriculture, and there are no pesticides necessary. In a market with a growing middle class seeking healthy foods, delivering from farm-to-door could let UrbanKisaan control quality and its margins.
Talyn Air: Two former SpaceX engineers have developed a long-range electric vertical take-off and landing (eVTOL) aircraft for passengers and cargo. The startup has created an electric fixed-wing aircraft that is caught mid-air with a custom winged drone during take offs and landings, an approach that its founders say give this aircraft three times the range of its competitors, at 350 miles.
BuildBuddy: Two ex-Googlers want to provide a “Google-style development environment” to all by building an open-source UI/feature set on top of Google’s Bazel software. The company says that their solution speeds up build times by up to 10x. It’s free for independent developers, with the price scaling from $4 per user to $49 per user depending on the size of the team and the features required.
Dataline: Meant to let websites gather analytics data from users who are using ad-blocking tools. Claiming that most ad-blocker users care mostly about display ads or cross-site tracking, the company says that first-party analytics gets hit as “collateral damage.” By acting as a “smart proxy” that runs on a sub-domain, Dataline avoids most ad-blocking systems (for now, presumably.)
Cortex: Many modern online software applications are powered by countless independent, purpose-focused tools — or “microservices.” Cortex monitors your app’s microservices to automatically flag the right person (hooking into Datadog/Slack/PagerDuty/etc.) when one breaks.
apitracker: Even if your website seems to be loading fine, the APIs you use to make it work might be having trouble, breaking things in not so obvious ways. Apitracker… tracks your APIs. It monitors the APIs you use, alerting you when one of them starts to fail and providing insights into their overall performance.
Freshpaint: Freshpaint’s “autotrack” system collects all pageviews/clicks/etc. across your site, allowing you to push it into tools like Google Analytics/Facebook Pixel etc. retroactively without requiring your dev team to make manual trackers for each event. The base plan is free for sites with fewer than 3,000 users and $300 for sites with up to 50,000 monthly users, after which point the pricing shifts to custom packaging.
Datree: Datree allows companies to set up rules and security policies for their codebase, and ensures those rules are followed before any code is merged. Charging $28 per developer (noting that it’s free for independent/open source projects), they’ve pulled in ~$230K in revenue to date. Find our previous coverage of Datree here.
fly.io: Deploys your app on servers that are physically closer to your users, decreasing latency and improving the user experience. If your app grows more popular in a certain city, Fly detects that and scales resources accordingly.
Sweeps: Sweeps claims that they can make your website 40% faster with one line of code, by more intelligently loading all of the third-party tools that a website is using. The team says that their tech not only improves speed but does so while improving SEO.
Orbiter: Orbiter is an automatic real-time monitoring and alert system integrated with Slack to ensure better customer service and revenue management.
Release: Product releases can be tricky. Release provides a staging management toolkit — it builds a staging environment each time there’s a pull request, allowing for faster/more collaborative development cycles.
Signadot: Signadot is monitoring and management software for the microservices that modern startups rely on to power their own applications and services, hopefully flagging issues before they become apparent to the end user.
Raycast: Raycast is a universal command bar for developers and many of the tools they use. Users can integrate apps including Jira, GitHub or Slack and take a Superhuman-like approach to completing forms and tasks. The team is pitching the tool as a way to help engineers get their non-engineering work done quickly.
Cotter: Cotter is building a phone number-based login platform that authenticates a user’s device in a workflow that the company’s founders say has the convenience of SMS-based OTP without the security issues. The startup is aiming to target customers in developing countries where email is less utilized and less convenient as a login.
ditto: Ditto’s founders are hoping to create the Figma for words, helping teams plan out more thoughtfully the copy they use to describe their products and workflows. The collaboration tool created by Stanford roommates Jolena Ma and Jessica Ouyang currently has 80+ different companies represented among their users.
Scout: A continuous integration and deployment toolkit for machine learning experiments inside a GitHub workflow.
ToDesktop: ToDesktop has designed a service to automate all of your desktop application publishing needs. It works with Windows, Mac and Linux and provides native installers, auto-updates, code signing and crash reports without the need for any infrastructure or configurations for developers.
DeepSource: DeepSource is a code review tool that allows developers to check for bug risks, anti-patterns, performance issues and security flaws in Python and Go.
Flowbot: Flowbot is a natural language, autocomplete search tool for coding in Python. It lets Python developers type in plain English when they can’t remember the exact function they’re thinking of, with Flowbot digging through documentation and considering the context to find the code it thinks you’re looking for.
PostHog: PostHog is a software service that lets developers understand how their users are actually working with their products. It’s a product analytics toolkit for open-source programmers.