This Week in AI Agents: Big Changes Are Coming

The week of January 26 through February 3, 2026, showed us that AI agents are becoming the next big thing in technology. Companies are rushing to build new tools and services around these smart computer programs that can work on their own.

Airtable's Bold New Move

Airtable, a company that helps people create software without knowing how to code, made a huge announcement this week. They launched something new called Superagent, which is a completely different kind of product from anything they've made before. The company has been around for 13 years, and this is their first stand-alone product.

What makes Superagent special is how it thinks about solving problems. When you ask Superagent a question, instead of just giving you an answer, it first makes a plan. It figures out what information it needs and what questions to ask. Then, instead of doing all the work one step at a time, it sends out a team of AI experts who all work at the same time. One AI expert might research money and numbers. Another might look at what competitors are doing. Another might check the news. They all work together, and then Superagent puts all the information together into a beautiful report.

The reports don't look like boring text documents. They look like something you'd see in a fancy newspaper, with colorful charts, maps, and numbers you can click on. Airtable's leader, Howie Liu, said that companies like the New York Times make beautiful information graphics, and now every person could have that quality of work for anything they need.

Why is Airtable doing this? The company had some tough times. Back in 2021, people thought the company was worth $11.7 billion. Today, it's worth about $4 billion on the open market. But Liu says the company still has money in the bank and doesn't need to borrow more. He's betting on Superagent to be the company's future. The team working on Superagent includes the people from a company called DeepSky that Airtable bought recently.

Mastercard Helps Businesses Use AI Agents

Mastercard, which is one of the biggest payment companies in the world, also announced their own AI agent service this week. They call it Agent Suite. But instead of making their own AI agent to do one type of work, Mastercard is helping other companies create and use AI agents.

Mastercard says that banks and stores need help moving into the age of AI agents. Mastercard will give them tools to build agents, test them, and put them to work. The company will also send experts to help each customer.

One example is a bank using an AI agent to recommend credit cards. When a customer looks at the bank's website, the AI agent can suggest a card that might be good for them, like a card for travelers or a card that saves money. The bank can test different suggestions to see which ones work best. Another example is a store using an AI agent to help shoppers. The AI agent can answer questions about products and suggest things the shopper might like.

Mastercard is building this with privacy and safety in mind. The Agent Suite will be available in the second quarter of 2026. This means customers might start using it around April or May.

Everyone Is Building AI Agents

This isn't just Airtable and Mastercard. The whole tech industry is racing to build AI agents. Companies like OpenAI, Notion, Harvey, and hundreds of others have added AI agents to their products. OpenAI started 2025 by launching new tools for building agents.

Experts are saying this is a huge trend. One expert said that AI is rapidly replacing the old way of writing computer code. Instead of people writing code, AI is writing it now. Multi-agent orchestration platforms like Claude-Flow and Zenflow are becoming essential tools. These platforms let multiple AI experts work together on problems.

According to Gartner, a company that watches tech trends, about 40 percent of all business software will use AI agents by 2026. Just a year ago, that number was less than 5 percent. That's a giant jump.

The Big Security Problem Nobody Is Talking About Enough

While companies are excited about AI agents, there's a serious problem that people are just starting to notice. AI agents need access to company data to do their jobs, but companies don't always keep good track of them.

Think about it this way: it's like having thousands of interns running around in your company with keys to all the important offices, and nobody knows exactly where they are. When security experts scan companies, they often find one to 17 AI agents for every single worker. Some companies didn't even know these AI agents existed.

The problem is that AI agents need special accounts and passwords to access company systems. These are called "agentic identities". But nobody has good systems yet to keep track of all these agent accounts like they do with human worker accounts. If someone tricks an AI agent with the wrong instructions, it could accidentally share secrets or break things.

One company discovered that someone could trick their AI agent into putting dangerous software on an employee's computer. Another problem is "shadow AI"—when someone uses a personal ChatGPT account or other AI tool without telling the company's IT department. That person might connect the AI to all the company's data sources, creating a super-connected agent.

Security experts are saying this is the biggest insider threat of 2026. They're calling for better ways to keep track of AI agents and who created them.