Artificial intelligence agents face one last, big obstacle

Integration of AI agents and apps via existing authorization protocols could also be the last major challenge

- Steven Rosenbush

The race to build useful artificial-intelligence agents that can perform complex actions for people is moving on to a new set of challenges.

The large language models at the core of these agents are good enough for many tasks. But there is a growing emphasis on connecting the LLMs inside agents to a plethora of tools that they will need to get their jobs done.

An advanced LLM might fail at complex multiplication, for example, while the cheapest, oldest model can ace the test if it has a calculator tool.

But there is another hurdle: Agents will need permission to access apps, APIs and websites if they are ever going to call an Uber or book a flight, the kind of expectation that has been established over the past year.

Humans type passwords or use facial and fingerprint recognition to sign into their accounts, but AI agents require new methods of authorization to address the intermediary role between humans and the services they want to use, according to Alex Salazar, chief executive of startup Arcade.dev.

The promise of agents