Agentic AI
,
Artificial Intelligence & Machine Learning
,
Identity & Access Management
Former AWS Exec Nancy Wang to Lead 1Password’s Agentic AI Security Strategy
Jennifer Lawinski •
January 12, 2026 Â Â
Nancy Wang, chief technology officer, 1Password (Image: 1Password)
As companies begin to advance agentic artificial intelligence projects from the pilot phase to production, new cybersecurity challenges are emerging as software acts on behalf of the enterprise, instead of humans.
See Also: Proof of Concept: Bot or Buyer? Identity Crisis in Retail
These agents represent a whole new class of machine identities that need to be secured, which is a major focus of password manager and identity security vendor 1Password. The company last week named former AWS executive Nancy Wang as chief technology officer to oversee the evolution of its platforms to manage new AI-driven workflows.
Wang, a former technology executive with Amazon Web Services’ data protection business, said called AI agents a new class of user identity that traditional security models were never intended to govern. “Agents are really their own class of identities,” she said in an interview with CIO.inc.
AI agents are dynamic and non-deterministic, so “it’s no longer this paradigm where input equals output,” Wang said. “Even if you give it the same input, the output might still change because agents are pieces of software that are built on top of LLMs.”
For CIOs facing pressure to advance AI pilots to production, this transition is creating governance challenges. AI agents are increasingly interacting with deterministic production systems such as databases, cloud consoles and production APIs, creating new risks. “If we assume agents are just another service account, we could protect ourselves from some of the risk, but not all of the risk,” Wang said.
Wang said her mandate over the next 12 to 18 months is to ensure 1Password becomes the default security platform enterprises turn to when protecting AI agents.
The company’s enterprise products are essentially credential vaults for human and service accounts, but the next step, Wang said, is expanding the vaults to include credentials for AI agents. Its “autofill” capabilities, which cryptographically inject credentials when they’re needed, will extend into AI workflows.
“It’s just in time,” Wang said. “That’s really where the magic’s going to be. You’re not just authenticating the agent, but you’re also authorizing the agent to take an action at that point in time, while all the while making sure that the keys of the kingdom for your enterprise, like your SSH keys or API keys to all the models are kept secure.”
In conversations with enterprise customers, Wang said CIOs have expressed concerns about data privacy and security, and how to manage internal AI literacy to ensure that employees use AI tools appropriately. So, keeping the user experience simple is one of her top priorities to support the company’s growth. “The secure way has to be the easy way,” she said. “Whether you’re a developer, or a business worker, or a knowledge worker, or a business worker, using 1Password just has to be a very simple experience.”
For example, 1Password recently integrated its products with AI-focused developer tools and browsers, including Cursor, BrowserBase, ChatGPT Atlas and Perplexity Comet. In Cursor, 1Password has built hooks that allow developers to identify risky credential practices, such as orphaned SSH keys or credentials stored in plain text, before AI-assisted code makes its way into production, she said.