Countless companies are researching AI, and suddenly data has become extremely valuable. A recent report from AWS shows that small and medium-sized enterprises (SMEs) that integrate data analytics into their business operations are more likely to use AI and outperform their competitors.
"The most surprising finding is that small and medium-sized enterprises (SMEs) that leverage data are performing exceptionally well financially," said Ben Schreiner, AWS's U.S. Small Business Innovation Leader.
A number of business tools have emerged on the market, leveraging large language models to analyze sales and expense data. Meanwhile, AI is increasingly appearing in scenarios involving text, images, and audio recordings, making this content increasingly valuable.
AI companies will reach agreements with news publishers, social media platforms, and image library operators to obtain licenses, acquire data, and develop general-purpose AI tools. Data owners can also use their data to train and enhance AI, which can then be used to serve employees and customers. Emails, historical financial reports, geographic data, legal documents, corporate forum posts, and customer service records can all serve as sources of information.
"The source material contains a wealth of knowledge (actionable information and content) that you can use as a foundation to develop all sorts of applications with unimaginable potential," said Edo Liberty, founder and CEO of vector database software developer Pinecone.
Vector databases store large amounts of documents or other files digitally, making it convenient to compare data documents. Whether searching for related materials, categorizing similar files, or providing suggestions based on past user interests, the speed is much faster. Vector databases can also be paired with AI to serve large language models, providing more accurate answers to user questions.
This is known as Retrieval-Augmented Generation (RAG), which takes generative AI to the next level, providing answers beyond the scope of general training data. Like other machine learning techniques, Retrieval-Augmented Generation relies on accurate and well-organized data.
"The quality of AI depends on the data. If there is no data, AI can't tell you anything," said Eilon Reshef, co-founder of revenue intelligence analytics firm Gong.
Gong's goal is simple: it extracts information from enterprise sales emails, call logs, and online interaction data to help enterprises analyze and manage sales activities.
To build more trustworthy AI, we must obtain cleaner and more reliable data. Being data-centric and ensuring the system complies with laws, regulations, and internal company rules is also crucial.
Walmart has developed a system called Element that helps businesses build reliable AI solutions that can span multiple cloud providers. The software acts as an facilitator, ensuring that businesses use data in compliance with regulations. It also monitors models to see if they are changing and whether they remain accurate.
Walmart executive Anil Madan stated, "When new data comes in, we need to monitor the model to see if it changes and continues to evolve. We don't want unnecessary biases in our AI systems. To achieve our goals, it's crucial to ensure data quality; data should help us use AI in a more responsible manner."
In the process of enterprise production, factors such as production, shipment, tariffs, and sanctions can all affect operations. Startup Altana has found a way to make money from this by analyzing data and helping enterprises avoid risks.
"Our models and systems can learn from the entire graph, find errors, check and correct them," said Peter Swartz, co-founder of Altana.
For B2B companies, they often serve multiple clients, who contribute a large amount of data. After learning from the data, AI can provide recommendations for specific clients.
Intuit has built a so-called email marketing automation platform that claims to convert email messages into revenue and provide marketing advice to businesses. The longer and more frequently a business uses the platform, the more reliable the AI-generated recommendations become.
"The deeper our understanding of small businesses, the more personalized our advice becomes," said Shivang Shah, chief architect at Intuit. "At the same time, the revenue engine will be more confident in its recommendations, telling businesses what kind of marketing campaigns they should undertake."
About a year ago, Intuit released a new platform called GenOS, which can help developers quickly develop AI tools.
It's easy to see that when AI is deeply integrated with enterprises and businesses, massive amounts of data become a goldmine, and AI data startups are likely to emerge in all industries. If you are aspiring to start a business, you may want to consider this direction.
