More than 40% of initiatives related to agent-based artificial intelligence may be shut down by 2027 due to high costs and a lack of tangible business value. Currently, most of these projects remain at the experimental stage and are unable to provide the necessary level of autonomy to solve complex business tasks.
This is reported by Business • Media
Reasons for Closing Agent-Based AI Projects
Analysts note that companies often rebrand old software solutions, presenting them as agent-based AI. In reality, only about 130 of the thousands of claimed products truly possess autonomous functions that allow systems to make decisions and take actions independently without constant human intervention. The rest of the solutions either have limited autonomy or lack it altogether.
According to Gartner analyst Anushree Verma, the current market is saturated with concepts that emerged during the wave of artificial intelligence popularity but do not demonstrate effective returns on investment.
“Most of these initiatives do not show a return on investment,” she stated.
Additionally, analysts point to the phenomenon of agent washing, where ordinary digital assistants or chatbots are promoted under the guise of agent-based AI, lacking true autonomy.
Prospects for the Development of Agent-Based AI
Despite the challenges, experts predict that in the long term, the implementation of agent-based AI will grow. By 2028, such systems are likely to participate in at least 15% of everyday work decisions, whereas in 2024, this figure was at 0%. Furthermore, it is forecasted that in three years, one-third of corporate software products will be based on agent-based AI solutions, compared to less than 1% today.
Major technology companies such as Google, Microsoft, Salesforce, and Oracle are actively investing billions of dollars in the development of agent-based AI systems, aiming to optimize costs and enhance the efficiency of business processes.
Earlier, it was reported that Donut Labs raised $7 million to create a crypto browser with built-in AI agents.