Despite an overall decline in startup investment, funding for AI rose sharply last yr. Capital for generative AI projects alone increased nearly fivefold from 2022 to 2023, reaching $25.2 billion at the top of December.
So it's not exactly surprising that AI startups dominated Y Combinator's Winter 2024 Demo Day.
The Y Combinator Winter 2024 cohort includes 86 AI startups, based on YC's official startup directory – almost twice as many as in winter 2023 and almost 3 times as many as in winter 2021. You can call it a bubble or exaggerate it, but AI is clearly the correct technology of the moment.
Like last yr, we checked out the newest Y Combinator cohort – the cohort that presented at Demo Day this week – and picked out among the more interesting AI startups. Everyone made the cut for various reasons. But at first glance, they stood out from the remaining, be it through their technology, their addressable market, or the background of their founders.
hazel
August Chen (ex-Palantir) and Elton Lossner (formerly of Boston Consulting Group) claim that the federal government contracting process is a hopeless failure.
Contracts are published on hundreds of various web sites and might contain lots of of pages of overlapping provisions. (The US federal government alone signs one estimated (greater than 11 million contracts per yr.) Responding to those offers can require the equivalent of entire business units, supported by outside consultants and law firms.
Chen and Lossner's solution is AI to automate the invention, drafting and compliance of presidency contracts. The couple, who met in college, calls it hazel.
Photo credit: hazel
Using Hazel, users can associate a possible contract, create a draft response based on their company's RFP and knowledge, create a checklist of tasks, and mechanically perform compliance checks.
Given the AI's propensity for hallucinations, I'm a bit of skeptical that Hazel's generated answers and checks are consistently accurate. But in the event that they get anywhere close, they may save enormous amounts of effort and time and provides smaller corporations a shot on the lots of of billions of dollars awarded in government contracts every year.
Andy AI
Home health nurses do quite a lot of paperwork. Tiantian Zha knows this well – she previously worked at Verily, Google's parent company alphabetShe worked within the biological sciences department, where she was involved in moonshots starting from personalized medicine to reducing mosquito-borne diseases.
In the course of her work, Zha discovered that documentation was a major time commitment for home care staff. It's a widespread problem – considered one of those studyNurses spend greater than a 3rd of their time on documentation, leading to less time spent on patient care contribute to burnout.
To reduce the documentation burden on caregivers, Zha co-founded Andy AI with Max Achterow, a former Apple Staff engineer. Andy is actually an AI-powered scribe that captures and transcribes the spoken details of a patient visit and creates electronic health records.
Photo credit: Andy AI
As with any AI-powered transcription tool, there may be Risk of bias – This signifies that the tool doesn’t work well for some nurses and patients depending on their accent and word selection. And from a competitive perspective, Andy isn't precisely the first of its kind to come back to market – competitors include DeepScribe, Heidi Health, Nabla and Amazon's AWS HealthScribe.
But as health care increasingly As business moves home, demand for apps like Andy AI appears to be increasing.
Particularly
If your experience with weather apps is anything like this reporter's, you've walked right into a rainstorm after blindly believing in predictions of clear, blue skies.
But it doesn't should be that way.
At least that's the premise of Precip, an AI-powered weather forecasting platform. Jesse Vollmar got here up with the thought after founding FarmLogs, a startup that sold crop management software. He teamed up with Sam Pierce Lolla and Michael Asher, the previous lead data scientist at FarmLogs Particularly a reality.
Photo credit: Particularly
Precip provides precipitation evaluation, resembling estimating the quantity of precipitation in a selected geographic area over the past few hours to days. Vollmar claims that Precip can generate “highly accurate” readings for any location within the U.S. right down to the mile (or two) and predict conditions as much as seven days prematurely.
What is the worth of rainfall metrics and warnings? Now, Vollmar says, farmers can use them to trace crop growth, construction crews can refer them to planning teams and utilities can use them to anticipate service interruptions. A shipping customer checks Precip each day to avoid poor driving conditions, Vollmar claims.
Of course, there is no such thing as a shortage of weather forecast apps. But AI like Precip's guarantees to make forecasts more accurate – if the AI is definitely definitely worth the money.
Maia
Claire Wiley began a couples coaching program while studying for her MBA at Wharton. This experience led her to explore a more technologically advanced approach to relationships and therapy, culminating on this Maia.
Maia – which Wiley co-founded with Ralph Ma, a former Google Research scientist, goals to empower couples to construct stronger relationships through AI-powered guidance. In Maia's apps for Android and iOS, couples send one another messages in a bunch chat and answer questions each day, resembling: B. what challenges they think they should overcome, past pain points, and lists of things they’re grateful for.
Photo credit: Maia
Maia plans to become profitable by charging for premium features like therapist-created programs and unlimited sending. (Typically, Maia limits texts between partners—a frustratingly arbitrary restriction should you ask me, but that's the way it is.)
Wiley and Ma, who each come from divorced households, say they worked with a relationship expert to design the Maia experience. The questions on my mind, nonetheless, are (1) how sound Maia's relationship science is and (2) can she stand out within the extremely crowded field of couples apps? We'll should wait and see.
Data curve
The AI models at the guts of generative AI apps like ChatGPT are trained on massive data sets, a combination of public and proprietary data from across the online, including e-books, social media posts and private blogs. However, a few of this data is legally and ethically problematic – not to say flawed other ways.
The distinct lack of information curation is the issue should you ask Serena Ge and Charley Lee.
Ge and Lee were co-founders Data curve, which provides “expert-quality” data for training generative AI models. In particular, this involves code data, which Ge and Lee say is especially difficult to acquire due to the expertise required for labeling for AI training and restrictive use licenses.
Photo credit: Data curve
Datacurve hosts a gamified annotation platform that pays engineers to unravel coding challenges and contributes to Datacurve's training datasets on the market. These datasets will be used to coach models for code optimization, code generation, debugging, UI design and more, say Ge and Lee.
That's definitely an interesting idea. But Datacurve's success will rely upon how well its data sets are curated – and whether it manages to motivate enough developers to proceed constructing on and improving them.
