VentureBeat
Follow
What MIT got wrong about AI agents: New G2 data shows they’re already driving enterprise ROI
A new G2 report reveals that AI agents are far more successful than previously believed, with nearly 60% of companies using them in production and less than 2% failing. This data contradicts earlier academic reports suggesting widespread AI project stagnation. The report surveyed over 1,300 B2B decision-makers, highlighting high satisfaction rates and significant cost savings, particularly in customer service, BI, and software development. Despite success, a "dead heat" exists between autonomous and hybrid models, with companies often balancing human oversight. Full autonomy is more accepted in low-risk workflows, like data management. Companies are also investing heavily in AI agents, with a majority planning to increase investment in the next year. Agentforce, a Salesforce product, is leading the market, but many aim to develop in-house tools. AI agents improve efficiency by overcoming human challenges like procrastination, enabling faster workflows with constant work. Security is a key concern, as nearly 40% reported security incidents. Explainability is key to building trust, and focusing on business problems before implementation. Trust in AI, like cloud computing, is a gradual process requiring open communication and data. Companies must prioritize measuring and retraining quickly following any agent mistakes. Therefore, approaching deployment correctly and mitigating potential trust issues can lead to success.
