The 11 next big things in AI and data innovations for 2024

The AI big bang that first caught the world’s attention with ChatGPT is sending energy and innovation out in many directions. Large language models aren’t just chatting and summarizing text: Now they’re also reasoning through problems with increasing autonomy. The text-only models of two years ago have become multi-model—capable of processing everything from audio and video to computer code and robotic movements. Big companies are deploying AI models and apps in new ways, and AI infrastructure startups are finding no shortage of pain points in that process to address. These companies deserve special attention for pushing these transformations toward real-life advances within business, science, and personal AI.

Bellwether
For applying AI to extreme weather event response
As climate change gives rise to more extreme weather, fires, and floods, the world needs new technologies, including AI, to understand them. Bellwether, which is being incubated at Alphabet’s X research lab, uses AI to process and analyze satellite imagery, then provides disaster response teams with the maps and insights they need to best deal with disasters on the ground. Since earth observation data is ever-changing, Bellwether developed advanced simulation techniques to unlock new insights into the effects of natural events on both human-made and natural structures. For example, its tools have been used to predict wildfires as well as to analyze hurricanes or flood zones, so that relief resources can be deployed more quickly.

Blackbird.AI
For creating a weapon against online misinformation
In an online world where bad actors muddy the waters with disinformation to the point where the truth can seem unknowable, netizens and brands need a sane fact-checker at their side. That’s what Blackbird.AI has built with Compass, a web app that lets users reality check suspicious claims made in social media posts, videos, links, and memes. When they paste a piece of content into the Compass tool, it checks it against thousands of trustworthy sources and generates an answer accompanied by footnotes and citation links. Businesses can use Compass to locate and counter false narratives about their brands. 

Databricks
For putting powerful AI models where the data lives
Enterprises are increasingly using AI models to look for insights within their proprietary data, from documents and video to transcripts of support calls. Hosting and enabling analytics on enterprise data within a single cloud is Databricks’ claim to fame. Databricks customers can train its DBRX AI model on their own data to get business-specific answers to questions. For example, a healthcare organization could use the model to look for patterns of disease across thousands of electronic patient records.

Enko Chem
For intelligently fighting food insecurity
Farmers around the world lose 30% to 40% of their crops to weeds, insects, and diseases that are spreading amidst changing climates. Enko Chem is using its Enkompass AI model to design novel, eco-friendly herbicides for agriculture. The model combs through proprietary encoded DNA libraries containing billions of chemical compounds to find the ideal molecules needed to target specific pests. It factors in safety and efficacy guardrails from the start, so that new herbicides can pass testing and regulatory hurdles and reach the market quickly.

Filecoin
For democratizing control of the web’s data
The majority of the web’s content and data is stored by just three cloud companies: Amazon Web Services, Microsoft Azure, and Google Cloud. The Filecoin peer-to-peer storeage network provides a decentralized alternative. The storage capacity comes from roughly 3,000 participating storage system operators, who get economic incentives to make sure files are stored reliably over time. While not a household name, Filecoin’s network is huge—big enough to hold 5.8 exabytes of data (the equivalent of 1.3 trillion copies of the Star Wars DVD). Among Filecoin’s clients: the Internet Archive, which stores more than a petabyte of data across the network.

HeyGen
For making video production less daunting
Businesses rely on videos as a primary way to communicate with customers, but they’ve had to put up with high production and translation costs to do so. That’s the pain point HeyGen’s AI app addresses. A multinational company executive might upload an employee announcement, then let the tool recreate a lifelike video avatar of her saying the same thing in a hundred different languages. Or she might just upload a video sample and the announcement script, and let the app do the rest. The company’s customers range from McDonald’s to the Tennis Channel.

Makersite
For bringing better environmental awareness to product design
Makersite’s platform taps  trusted supply chain databases and uses AI to give designers a clear look at the carbon footprint of every material and component they use. Armed with that knowledge, they can quickly make changes to comprehensively manage sustainability, cost, and compliance. Makersite’s biggest client, Microsoft, used the platform to reduce the carbon footprint of its Surface Pro 10 by 28% within two years.

Nixtla
For turning data into smarter predictions
Time-series AI models process large amounts of time-stamped data to predict future events and business conditions. Nixtla (which means “time” in the Aztec language) offers a time-series model called TimeGPT that even small companies can access via an API. The model was trained on more than 100 billion pieces of financial, weather, energy, and web data, and customers can fine-tune it using their own data. The company offers an open-source version of its model, as well as a version for Microsoft Azure environments.

Qualcomm
For powering AI models on mobile devices
Today the vast majority of AI models run on powerful servers within data centers. But as models become more efficient and chips become more powerful, it will become common for phones and other mobile devices to contain their own AI brains. Qualcomm’s Snapdragon 8 Gen 3 chips are already enabling new AI features on more than 100 devices, including phones from manufacturers such as Samsung, Honor, and Xiaomi. Features such as live translation that run on-device are lightning fast and privacy conscious, since no data is sent to the cloud.

Runway
For pushing the envelope in AI-generated video
AI-generated video made large leaps in 2024, both in quality and in the maximum duration of generated clips, and the New York-based video generation startup Runway has remained a pacesetter in that progress. In June 2024, it released its new Gen-3 Alpha model, which offers noticeably better video quality than previous versions, and smarter tools for controlling the look of the video. The tool now generates single clips of up to 10 seconds, with options for add-ons of up to 40 seconds. Runway also recently announced a new Gen-3 tool called Act One that creates animated character performances from a simple video of an actor performing.

Sana AI
For applying AI to even disorganized data
Enterprises are hurrying to put large language models to work, but many have stumbled because of messy, scattered data. Enter Sana AI, a “universal knowledge assistant” that mediates between a company’s data and whatever large language models it uses, allowing workers to access unstructured data from one place using plain language chat or search. The system can pull data from a company’s apps or access conversations held during meetings via video conference call transcripts. Among its uses are analyzing documents, drafting reports, and automating repetitive tasks.

The companies behind these technologies are among the honorees in Fast Company’s Next Big Things in Tech awards for 2024. See a full list of all the winners across all categories and read more about the methodology behind the selection process.

https://www.fastcompany.com/91207410/fast-company-next-big-things-ai-data-2024?partner=rss&utm_source=rss&utm_medium=feed&utm_campaign=rss+fastcompany&utm_content=rss