AI Agents: ServiceNow trains them smarter with synthetic data
·2 min read·Intermediate
“
Training AI with real data is a nightmare of privacy and scarcity. ServiceNow found a fix: make it learn with fake but realistic data.
In 30 seconds
01ServiceNow AI developed AutoSynthData to generate synthetic training data.
02This fake data helps enterprise AI agents learn faster and perform better.
→
💡
What this means for you
For the average person, this means enterprise virtual assistants will get smarter and more helpful, but without your personal data ending up in their training. Fewer frustrations when talking to a bot, in short.
Tired of AI agents living in the cloud, far away? A new open source project brings them straight to your home computer. Less cloud, more control, at least on paper.
·2 min·2·Intermediate
03It enables effective virtual assistants without using sensitive information.
0101
Why is real data a headache for AI?
Training an for the enterprise world is a real pain. Imagine having to teach a bot how to handle complaints or HR queries using real data. Sure, because finding "real" and "clean" data is like looking for a needle in a spreadsheet haystack, and one full of super sensitive info you can't show anyone. A real mess, in short.
Specific data is needed for each sector, each company, each individual case. Often, this data doesn't exist in sufficient quantities or is too delicate to be used freely. And who would've thought artificial intelligence would learn better with a bit of healthy fiction? Apparently, not just politicians.
0202
How does AutoSynthData's synthetic data work?
ServiceNow AI developed AutoSynthData precisely for this: to generate synthetic training data for enterprise agents. In practice, it's like creating a parallel universe of conversations, scenarios, and interactions, without touching a single piece of real, sensitive company data. This data is fake, but designed to be indistinguishable from real data, at least for the AI.
📬 Enjoying this article?
Get the best AI news every week, straight to your inbox.
AutoSynthData uses large language models (LLMs) to invent these scenarios. It generates dialogues, problem descriptions, and solutions, all tailored for the agent being trained. It's a bit like giving your student a math textbook full of made-up problems that look real and prepare them for the actual exam. It works, apparently.
0303
What changes for enterprise AI agents?
With AutoSynthData, AI agents can learn much faster and with greater accuracy. Imagine a virtual assistant answering IT or HR questions: with synthetic data, it can be trained to handle thousands of different scenarios without a single employee having to provide their personal data. Fewer privacy headaches, more efficiency for everyone.
This approach speeds up the development of AI agents, making them more robust and reliable from the get-go. It's not just a time saver, but also a resource saver, as collecting and anonymizing real data is a lengthy and expensive process. In short, a big step forward for anyone wanting AI that works without creating more problems.
NUOVO
