Regardless of an total decline in startup investing, funding for AI surged prior to now yr. Capital towards generative AI ventures alone practically octupled from 2022 to 2023, reaching $25.2 billion towards the tail finish of December.

So it’s not precisely shocking that AI startups dominated at Y Combinator’s Winter 2024 Demo Day.

The Y Combinator Winter 2024 cohort has 86 AI startups, in keeping with YC’s official startup listing — practically double the quantity from the Winter 2023 batch and near triple the quantity from Winter 2021. Name it a bubble or overhyped, however clearly, AI is the tech of the second.

As we did final yr, we went by way of the most recent Y Combinator cohort — the cohort presenting throughout this week’s Demo Day — and picked out among the extra fascinating AI startups. Every made the reduce for various causes. However at a baseline, they stood out among the many relaxation, whether or not for his or her expertise, addressable market or founders’ backgrounds.

Hazel

August Chen (ex-Palantir) and Elton Lossner (ex-Boston Consulting Group) assert that the federal government contracting course of is hopelessly damaged.

Contracts are posted to hundreds of various web sites and might embrace a whole bunch of pages of overlapping laws. (The U.S. federal authorities alone indicators an estimated 11 million+ contracts a yr.)  Responding to those bids can take the equal of entire enterprise divisions, supported by exterior consultants and regulation corporations.

Chen’s and Lossner’s resolution is to make use of AI to automate the federal government contracting discovery, drafting and compliance course of. The pair — who met in faculty — name it Hazel.

Utilizing Hazel, customers can get matched to a possible contract, generate a draft response primarily based on the RFP (request for proposal) and their firm’s information, create a guidelines of to-dos and routinely run compliance checks.

Given AI’s tendency to hallucinate, I’m a bit skeptical that Hazel’s generated responses and checks shall be persistently correct. However, in the event that they’re even shut, they may save an infinite quantity of effort and time, enabling smaller corporations a shot on the a whole bunch of billions of {dollars}’ price of presidency contracts issued annually.

Andy AI

House nurses take care of numerous paperwork. Tiantian Zha is aware of this effectively — she beforehand labored at Verily, the life sciences division of Google mother or father firm Alphabet, the place she was concerned in moonshots starting from customized medication to decreasing mosquito-borne illnesses.

In the midst of her work, Zha discovered that documentation was a significant time sink for at-home nurses. It’s a widespread situation — in keeping with one research, nurses spend over a 3rd of their time on documentation, chopping into time spent on affected person care and contributing to burnout.

To assist ease the documentation burden for nurses, Zha co-founded Andy AI with Max Akhterov, a former Apple workers engineer. Andy is basically an AI-powered scribe, capturing and transcribing the spoken particulars of a affected person go to and producing digital well being information.

As with every AI-powered transcription device, there’s threat of bias — that’s, the device not working effectively for some nurses and sufferers relying on their accents and phrases decisions. And, from a aggressive standpoint, Andy isn’t precisely the primary of its form to market — rivals embrace DeepScribe, Heidi Well being, Nabla and Amazon’s AWS HealthScribe.

However as healthcare more and more shifts to house, the demand for apps like Andy AI appears poised to extend.

Precip

In case your expertise with climate apps is something like this reporter’s, you’ve been caught in a rainstorm after blindly believing predictions of clear blue skies.

However it doesn’t need to be this fashion.

Not less than, that’s the premise of Precip, an AI-powered climate forecasting platform. Jesse Vollmar had the thought after founding FarmLogs, a startup that bought crop administration software program. He teamed up with Sam Pierce Lolla and Michael Asher, beforehand FarmLogs’ lead knowledge scientist, to make Precip a actuality.

Precip delivers analytics on precipitation — for instance, estimating the quantity of rainfall in a given geographic space over the previous a number of hours to days. Vollmar makes the declare that Precip can generate “high-precision” metrics for any location within the U.S. all the way down to the kilometer (or two), forecasting circumstances as much as seven days forward.

So what’s the worth of precipitation metrics and alerts? Nicely, Vollmar says that farmers can use them to trace crop progress, building crews can reference them to schedule crews, and utilities can faucet them to anticipate service disruptions. One transportation buyer checks Precip every day to keep away from dangerous driving circumstances, Vollmar claims.

After all, there’s no scarcity of climate prediction apps. However AI like Precip’s guarantees to make forecasts extra correct — if the AI is price its salt, certainly.

Maia

Claire Wiley launched a {couples} teaching program whereas finding out for her MBA at Wharton. The expertise led her to research a extra tech-forward method to relationships and remedy, which culminated in Maia.

Maia — which Wiley co-founded with Ralph Ma, a former Google analysis scientist — goals to empower {couples} to construct stronger relationships by way of AI-powered steerage. In Maia’s apps for Android and iOS, {couples} message one another in a bunch chat and reply every day questions like what they view as challenges to beat, previous ache factors and lists of issues that they’re grateful for.

Maia plans to become profitable by charging for premium options corresponding to applications crafted by therapists and limitless messaging. (Maia presently caps texts between companions — a frustratingly arbitrary limitation if you happen to ask me, however so it goes.)

Wiley and Ma, each of whom come from divorced households, say that they labored with a relationship professional to craft the Maia expertise. The questions in my thoughts, although, are (1) how sound is Maia’s relationship science and (2) can it stand out within the exceptionally crowded discipline of {couples}’ apps? We’ll have to attend to see.

Datacurve

The AI fashions on the coronary heart of generative AI apps like ChatGPT are skilled on monumental datasets, mixes of public and proprietary knowledge from across the internet, together with ebooks, social media posts and private blogs. However a few of this knowledge is legally and ethically problematic — to not point out flawed in different methods.

The distinct lack of knowledge curation is the issue, if you happen to ask Serena Ge and Charley Lee.

Ge and Lee co-founded Datacurve, which supplies “expert-quality” knowledge for coaching generative AI fashions. It’s particularly code knowledge, which Ge and Lee say is particularly arduous to acquire due to the experience essential to label it for AI coaching and restrictive utilization licenses.

Datacurve hosts a gamified annotation platform that pays engineers to resolve coding challenges, which contributes to Datacurve’s for-sale coaching datasets. These datasets can be utilized to coach fashions for code optimization, code era, debugging, UI design and extra, Ge and Lee say.

It’s an fascinating thought. However Datacurve’s success will rely on simply how well-curated its datasets are — and whether or not it’s capable of incentivize sufficient devs to proceed constructing on and bettering them.