\n
![](\"https:\/\/images.ctfassets.net\/jdtwqhzvc2n1\/5mIuAF6rRa5YJq9dkzNaRb\/3bb2774412459aecbe9e0718901209d7\/u7277289442_AI_agents_and_human_workers_sit_side-by-side_at_com_4680144f-3cfb-4bf3-9663-98adb31cb8e2.png\")
<\/p>\n
\n<\/p>\n
Check your research, MIT: 95% of AI projects<\/u> aren\u2019t failing \u2014 far from it.<\/p>\n
According to new data from G2, nearly 60% of companies already have AI agents in production, and fewer than 2% actually fail once deployed. That paints a very different picture from recent academic forecasts suggesting widespread AI project stagnation.<\/p>\n
As one of the world\u2019s largest crowdsourced software review platforms, G2\u2019s dataset reflects real-world adoption trends \u2014 which show that AI agents are proving far more durable and \u201csticky\u201d than early generative AI pilots.<\/p>\n
\u201cOur report\u2019s really pointing out that agentic is a different beast when it comes to AI with respect to failure or success,\u201d Tim Sanders, G2\u2019s head of research, told VentureBeat.\u00a0<\/p>\n
Sanders points out that the now oft-referenced MIT study<\/u>, released in July, only considered gen AI custom projects, Sanders argues, and many media outlets generalized that to AI failing 95% of the time. He points out that university researchers analyzed public announcements, rather than closed-loop data. If companies didn\u2019t announce a P&L impact, their projects were considered a failure \u2014 even if they really weren\u2019t.\u00a0<\/p>\n
G2\u2019s 2025 AI Agents Insights Report<\/u>, by contrast, surveyed more than 1,300 B2B decision-makers, finding that:\u00a0<\/p>\n 57% of companies have agents in production and 70% say agents are \u201ccore to operations\u201d;<\/p>\n<\/li>\n 83% of are satisfied with agent performance;<\/p>\n<\/li>\n Enterprises are now investing an average of $1 million-plus annually, with 1 in 4 spending $5 million-plus;\u00a0<\/p>\n<\/li>\n 9 out of 10 plan to increase that investment over the next 12 months;\u00a0<\/p>\n<\/li>\n Organizations have seen 40% cost savings, 23% faster workflows, and 1 in 3 report 50%-plus speed gains, particularly in marketing and saless;<\/p>\n<\/li>\n Nearly 90% of study participants reported higher employee satisfaction in departments where agents were deployed.<\/p>\n<\/li>\n<\/ul>\n The leading use cases for AI agents? Customer service, business intelligence (BI) and software development.\u00a0<\/p>\n Interestingly, G2 found a \u201csurprising number\u201d (about 1 in 3) of what Sanders calls \u2018let it rip\u2019 organizations.\u00a0<\/p>\n \u201cThey basically allowed the agent to do a task and then they would either roll it back immediately if it was a bad action, or do QA so that they could retract the bad actions very, very quickly,\u201d he explained.\u00a0<\/p>\n At the same time, though, agent programs with a human in the loop were twice as likely to deliver cost savings \u2014 75% or more \u2014 than fully autonomous agent strategies.<\/p>\n This reflects what Sanders called a \u201cdead heat\u201d between \u2018let it rip\u2019 organizations and \u2018leave some human gates\u2019 organizations. \u201cThere's going to be a human in the loop years from now,\u201d he said. \u201cOver half of our respondents told us there's more human oversight than we expected.\u201d\u00a0<\/p>\n However, nearly half of IT buyers are comfortable with granting agents full autonomy in low-risk workflows such as data remediation or data pipeline management. Meanwhile, think of BI and research as prep work, Sanders said; agents gather information in the background to prepare humans to make last passes and final decisions.\u00a0<\/p>\n A classic example of this is a mortgage loan, Sanders noted: Agents do everything right up until the human analyzes their findings and yay or nays the loan.\u00a0<\/p>\n If there are mistakes, they're in the background. \u201cIt just doesn't publish on your behalf and put your name on it,\u201d said Sanders. \u201cSo as a result, you trust it more. You use it more.\u201d\u00a0<\/p>\n When it comes to specific deployment methods, Salesforce's Agentforce \u201cis winning\u201d over ready-made agents and in-house builds, taking up 38% of all market share, Sanders reported. However, many organizations seem to be going hybrid with a goal to eventually stand up in-house tools.\u00a0<\/p>\n Then, because they want a trusted source of data, \u201cthey're going to crystallize around Microsoft, ServiceNow, Salesforce, companies with a real system of record,\u201d he predicted.\u00a0<\/p>\n Why are agents (in some instances at least) so much better than humans? Sanders pointed to a concept called Parkinson's Law<\/u>, which states that \u2018work expands so as to fill the time available for its completion.\u2019<\/p>\n \u201cIndividual productivity doesn't lead to organizational productivity because humans are only really driven by deadlines,\u201d said Sanders. When organizations looked at gen AI projects, they didn\u2019t move the goal posts; the deadlines didn\u2019t change.\u00a0<\/p>\n \u201cThe only way that you fix that is to either move the goal post up or deal with non-humans, because non-humans aren't subject to Parkinson's Law,\u201d he said, pointing out that they\u2019re not afflicted with \u201cthe human procrastination syndrome.\u201d<\/p>\n Agents don't take breaks. They don't get distracted. \u201cThey just grind so you don't have to change the deadlines,\u201d said Sanders.\u00a0<\/p>\n \u201cIf you focus on faster and faster QA cycles that may even be automated, you fix your agents faster than you fix your humans.\u201d\u00a0<\/p>\n Still, Sanders sees AI following the cloud when it comes to trust: He remembers in 2007 when everyone was quick to deploy cloud tools; then by 2009 or 2010, \u201cthere was kind of a trough of trust.\u201d\u00a0<\/p>\n Mix this in with security concerns: 39% of all respondents to G2\u2019s survey said they\u2019d experienced a security incident<\/u> since deploying AI; 25% of the time, it was severe. Sanders emphasized that companies must think about measuring in milliseconds how quickly an agent can be retrained to never repeat a bad action again.\u00a0<\/p>\n Always include IT operations in AI deployments, he advised. They know what went wrong with gen AI and robotic process automation (RPA) and can get to the bottom of explainability, which leads to a lot more trust.\u00a0<\/p>\n On the flip side, though: Don't blindly trust vendors. In fact, only half of respondents said they did; Sanders noted that the No. 1 trust signal is agent explainability. \u201cIn qualitative interviews, we were told over and over again, if you [a vendor] can't explain it, you can't deploy it and manage it.\u201d\u00a0<\/p>\n It\u2019s also critical to begin with the business problem and work backwards, he advised: Don't buy agents, then look for a proof of concept. If leaders apply agents to the biggest pain points, internal users will be more forgiving when incidents occur, and more willing to iterate, therefore building up their skillsets.\u00a0<\/p>\n \u201cPeople still don't trust the cloud, they definitely don't trust gen AI, they might not trust agents until they experience it, and then the game changes,\u201d said Sanders. \u201cTrust arrives on a mule \u2014 you don\u2019t just get forgiveness.\u201d<\/p>\n\n
AI agents aren't deadline-driven<\/h2>\n
Start with business problems, understand that trust is a slow build<\/h2>\n