Big tech bets big on autonomous computers

Major tech companies are advancing plans for computers that can manage complex tasks without constant user input. Tools like Alexa and Siri have handled basic functions such as setting alarms or playing music. Recent developments show this could change, with new hardware, software, and chips enabling AI agents to take on more responsibilities.

New hardware reduces dependence on manual commands

Nvidia introduced a chip named RTX Spark for Windows laptops. The component integrates graphics, computing, and networking with increased memory, allowing AI agents to operate locally instead of relying on cloud connections. Dell, HP, and Lenovo plan to release devices using the chip this fall.

Microsoft is updating Windows to support more autonomous features. The company launched an AI agent called Scout for Microsoft 365, which monitors emails and work chats across cloud storage, personal computers, and web apps like Outlook and Teams. The tool uses technology from OpenClaw, an AI assistant popular among developers.

Google is testing similar features. Its upcoming Googlebooks software can suggest actions when users hover over items on their screens. For example, pointing a mouse at a date in an email could prompt the system to set up a meeting.

Shift from task-specific tools to broader AI agents

Earlier digital assistants handled individual tasks like booking a taxi or placing an order. However, large language models like ChatGPT have changed the setting. OpenClaw, for instance, can run programs and complete requests without constant user input. Some developers now use it to manage research tasks on home computers, checking in via WhatsApp or Telegram.

“Things are quite different now because more people have become used to tools like ChatGPT or Gemini,” said David Naranjo of market research firm Counterpoint Research. “That’s created a new expectation for how computing should work.”

Nvidia CEO Jensen Huang demonstrated the potential of the new hardware at a press conference, showing how a laptop could use AI agents to assist with 3D modeling tasks while designing a house. The system worked between apps, reducing the need for manual input.

Challenges remain before widespread adoption

Despite these advances, experts caution that mass adoption is still years away. New laptops with the required hardware will likely be expensive, and consumer demand may not justify the cost. Companies like Anthropic and OpenAI are rapidly updating their AI tools, but keeping up with hardware changes could be a barrier.

“It’s not yet become indispensable,” said Naranjo. “That’s where the challenge exists for Nvidia and Microsoft and others.”

Another hurdle is trust. If an AI agent buys concert tickets and misinterprets a budget, users could face unexpected costs. “There’s a whole host of issues that need to be resolved before this becomes mass market,” said Jitesh Ubrani of International Data Corporation. “But are we on the right track? Yes.”

For now, the technology is more appealing to businesses than individual consumers. Local processing can reduce security risks by avoiding cloud data transfers, and companies may benefit from lower costs. But for average users, the shift to computers that control themselves remains a work in progress.