AI Agent Statistics 2026
This page collects published figures on AI agents in business: how many organizations are piloting or running agents in production, what they spend, what value they report, how agents score on task completion benchmarks, how much executives trust them, and how many projects analysts expect to fail. Most surveys cover large enterprises in the United States or global samples of companies with more than $1 billion in revenue. Data runs from 2024 benchmarks to KPMG's third-quarter 2026 survey.
By the DBotMaker research desk · Updated November 6, 2026 · 51 numbers drawn from 19 publishers and reports
Short answer
Most large companies now run AI agents in some form, but few have scaled them: KPMG's Q3 2026 AI Pulse found 62% of US organizations building, testing or deploying agents, while Capgemini Research Institute found only 2% had fully scaled deployment. Gartner predicts over 40% of agentic AI projects will be canceled by the end of 2027, and on benchmarks the best agents still fail roughly 1 in 3 OSWorld computer tasks, according to the Stanford AI Index 2026.
Headline numbers
- 62%62% of large US organizations surveyed by KPMG in the third quarter of 2026 were building, testing or deploying AI agents, up from 53% one quarter earlier.KPMG
- 2%Only 2% of large organizations surveyed by Capgemini Research Institute in April 2025 had fully scaled their deployment of AI agents.Capgemini Research Institute
- $207 millionUS organizations in KPMG's Q1 2026 AI Pulse projected average AI spending of $207 million over the next 12 months, nearly double the level a year earlier.KPMG
- 66%AI agents rose from 12% to about 66% task success on OSWorld, a benchmark of real computer tasks, but still fail roughly 1 in 3 attempts, according to the Stanford AI Index 2026.Stanford HAI
- 30%The best agent tested on TheAgentCompany, a simulated software company benchmark from Carnegie Mellon University researchers, completed 30% of tasks autonomously.Carnegie Mellon University
- 27%Trust in fully autonomous AI agents among large organizations fell from 43% to 27% in a year, Capgemini Research Institute found in 2025.Capgemini Research Institute
- 40%Gartner predicts over 40% of agentic AI projects will be canceled by the end of 2027 because of escalating costs, unclear business value or inadequate risk controls.Gartner
- 40%Gartner predicts up to 40% of enterprise applications will include integrated task-specific AI agents by 2026, up from less than 5% in 2025.Gartner
Reading the numbers
Adoption numbers depend heavily on how a survey defines an agent. Among US companies adopting agents, PwC found 35% using them broadly and 17% in almost all workflows in 2025, and Google Cloud put active use at 52% of executives at firms already running generative AI. Narrower questions give smaller answers: Capgemini Research Institute found 14% of large organizations had begun implementation and only 2% had fully scaled, and Gartner found only 15% of IT application leaders considering, piloting or deploying fully autonomous agents. KPMG's quarterly tracking shows the direction clearly, with the share of US organizations building, testing or deploying agents rising from 53% to 62% in one quarter of 2026.
Money is moving faster than results. KPMG respondents projected average AI spending of $207 million over the next 12 months in early 2026, and 88% of PwC respondents expected to raise AI budgets because of agentic AI. At the same time, KPMG found 74% of leaders now require cost reviews before approving AI projects and 43% enforce token or compute budgets, a sign that finance teams are pushing back on open-ended agent spending. Deloitte found only 25% of companies had moved 40% or more of their AI pilots into production.
Benchmarks explain part of the gap. The Stanford AI Index 2026 reports agents rising from 12% to about 66% task success on OSWorld, and METR measures the length of tasks agents can finish doubling about every 7 months. But in realistic business settings success rates are lower: Carnegie Mellon's TheAgentCompany benchmark found the best agent completed 30% of workplace tasks, and Salesforce AI Research found leading agents dropping to about 35% success in multi-turn CRM conversations. METR's 2025 trial even found experienced developers took 19% longer with AI tools.
Trust and governance are the main brakes. Capgemini found trust in fully autonomous agents fell from 43% to 27% in a year, Gartner found 74% of IT application leaders see agents as a new attack vector, and Deloitte found only 21% of companies have a mature governance model for agents. Gartner expects over 40% of agentic AI projects to be canceled by the end of 2027, even as it forecasts 40% of enterprise applications including task-specific agents by the end of 2026.
How many companies use AI agents
- 1.
62% of large US organizations surveyed by KPMG in the third quarter of 2026 were building, testing or deploying AI agents, up from 53% one quarter earlier.
KPMG, KPMG AI Quarterly Pulse Survey (Oct 2026)
Exact wording at the source
62% of organizations are currently building, testing, or deploying AI agents, up from 53% last quarter.
- 2.
25% of organizations in KPMG's Q3 2026 survey had deployed multi-agent systems, compared with 6% across the two previous quarters.
KPMG, KPMG AI Quarterly Pulse Survey (Oct 2026)
Exact wording at the source
25% of organizations have deployed multi-agent systems, up from 6% across the previous two quarters.
- 3.
35% of US executives whose companies were adopting AI agents told PwC in 2025 they were doing so broadly, and another 17% said agents were fully adopted in almost all workflows.
PwC, PwC's AI Agent Survey (May 2025)
Exact wording at the source
Of those companies adopting AI agents, 35% say they’re doing so broadly, while another 17% say AI agents are being fully adopted in almost all workflows and functions.
- 4.
68% of US executives at companies adopting AI agents told PwC in 2025 that half or fewer of their employees interact with agents in their everyday work.
PwC, PwC's AI Agent Survey (May 2025)
Exact wording at the source
Still, most (68%) report that half or fewer of their employees interact with agents in their everyday work.
- 5.
52% of 3,466 senior leaders at enterprises already using generative AI said their organization was actively using AI agents in 2025, and 39% said it had launched more than ten.
Google Cloud, Google Cloud Study Reveals 52% of Executives Say Their Organizations Have Deployed AI Agents (Sep 2025)
Exact wording at the source
More than half (52%) of executives report their organization is actively using AI agents, with 39% reporting their company has launched more than ten.
- 6.
Only 2% of large organizations surveyed by Capgemini Research Institute in April 2025 had fully scaled their deployment of AI agents.
Capgemini Research Institute, Rise of agentic AI: How trust is the key to human-AI collaboration (Jul 2025)
Exact wording at the source
Agentic AI is poised to deliver up to $450 billion in economic value by 2028 yet, despite strong momentum, only 2% of organizations have fully scaled deployment and trust in AI agents is declining .
- 7.
Among 1,500 executives at organizations with more than $1 billion in revenue, nearly a quarter had launched agent pilots and 14% had begun implementation in 2025.
Capgemini Research Institute, Rise of agentic AI: How trust is the key to human-AI collaboration (Jul 2025)
Exact wording at the source
While nearly a quarter have already launched pilots and a small number have begun implementation (14%), the majority remain in planning mode.
- 8.
57.3% of more than 1,300 professionals in LangChain's 2025 agent engineering survey had agents running in production, up from 51% a year earlier.
LangChain, State of Agent Engineering (Dec 2025)
Exact wording at the source
More than half of respondents surveyed (57.3%) now have agents running in production environments, with another 30.4% actively developing agents with concrete plans to deploy them.
- 9.
Only 25% of 3,235 business and IT leaders surveyed by Deloitte in August and September 2025 had moved 40% or more of their AI pilots into production.
Deloitte, State of AI in the Enterprise 2026 (Jan 2026)
Exact wording at the source
Deloitte’s survey found that only 25% of respondents have moved 40% or more of their AI pilots into production.
- 10.
85% of companies in Deloitte's 2026 State of AI survey expected to customize AI agents to fit the needs of their business.
Deloitte, State of AI in the Enterprise 2026 (Jan 2026)
Exact wording at the source
Autonomous AI agents are racing into the enterprise, transforming AI from a source of information and insights into a system that can do real work with 85% of companies expecting to customize agents to fit the unique needs of their business.
- 11.
81% of business leaders in Microsoft's 2025 Work Trend Index expected agents to be moderately or extensively integrated into their company's AI strategy within 12 to 18 months.
Microsoft, 2025: The Year the Frontier Firm Is Born (Work Trend Index) (Apr 2025)
Exact wording at the source
82% of leaders say this is a pivotal year to rethink key aspects of strategy and operations, and 81% say they expect agents to be moderately or extensively integrated into their company’s AI strategy in the next 12–18 months.
- 12.
Only 15% of 360 IT application leaders surveyed by Gartner in 2025 were considering, piloting or deploying fully autonomous AI agents.
Gartner, Gartner Survey Finds Just 15% of IT Application Leaders Are Considering, Piloting, or Deploying Fully Autonomous AI Agents (Sep 2025)
Exact wording at the source
Only 15% of IT application leaders said they are currently considering, piloting, or deploying fully autonomous AI agents (goal driven AI tools that do not require human oversight)
Chart
Agent builders with agents in production or in development
Share of more than 1,300 professionals surveyed by LangChain, 2025 and 2024 surveys. Respondents skew toward developers already working with agents.
- In production, 2024 survey51%
- In production, 2025 survey57.3%
- Developing with plans to deploy, 202530.4%
Source: LangChain, State of Agent Engineering
AI agent budgets and spending
- 13.
US organizations in KPMG's Q1 2026 AI Pulse projected average AI spending of $207 million over the next 12 months, nearly double the level a year earlier.
KPMG, KPMG AI Quarterly Pulse Survey (Apr 2026)
Exact wording at the source
Capital continues to flow into AI, with organizations projecting average AI spending of $207 million over the next 12 months, nearly double figures from the same period last year, according to the KPMG US Q1 AI Quarterly Pulse.
- 14.
74% of leaders in KPMG's Q3 2026 survey required cost reviews during AI project approval, up from 61% the previous quarter.
KPMG, KPMG AI Quarterly Pulse Survey (Oct 2026)
Exact wording at the source
74% of leaders require cost reviews during project approval, up from 61% last quarter.
- 15.
43% of organizations in KPMG's Q3 2026 survey enforced token or compute usage budgets to control AI spending.
KPMG, KPMG AI Quarterly Pulse Survey (Oct 2026)
Exact wording at the source
43% enforce token or compute usage budgets to manage spending.
- 16.
88% of US senior executives surveyed by PwC in 2025 said their team or function planned to increase AI budgets in the next 12 months because of agentic AI.
PwC, PwC's AI Agent Survey (May 2025)
Exact wording at the source
Of the 300 senior executives in our May 2025 survey, 88% say their team or business function plans to increase AI-related budgets in the next 12 months due to agentic AI.
- 17.
13% of executives in Google Cloud's 2025 ROI of AI study were agentic early adopters dedicating at least 50% of their future AI budget to AI agents.
Google Cloud, Google Cloud Study Reveals 52% of Executives Say Their Organizations Have Deployed AI Agents (Sep 2025)
Exact wording at the source
A distinct group of "agentic AI early adopters," representing 13% of executives surveyed, indicate their organizations are dedicating at least 50% of their future AI budget to AI agents and already have agents deeply embedded across operations.
- 18.
19% of 3,412 attendees in a January 2025 Gartner webinar poll said their organization had made significant investments in agentic AI.
Gartner, Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027 (Jun 2025)
Exact wording at the source
According to a January 2025 Gartner poll of 3,412 webinar attendees, 19% said their organization had made significant investments in agentic AI, 42% had made conservative investments, 8% no investments, with the remaining 31% taking a wait and see approach or are unsure.
- 19.
Gartner's best case projection has agentic AI driving about 30% of enterprise application software revenue by 2035, more than $450 billion, up from 2% in 2025.
Gartner, Gartner Predicts 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026 (Aug 2025)
Exact wording at the source
Gartner’s best case scenario projection predicts that agentic AI could drive approximately 30% of enterprise application software revenue by 2035, surpassing $450 billion, up from 2% in 2025.
Chart
Organizations' investment in agentic AI, January 2025
Gartner poll of 3,412 webinar attendees, January 2025. Not a random sample of organizations.
- Conservative investments42%
- Wait and see or unsure31%
- Significant investments19%
- No investments8%
Source: Gartner, Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027
Reported value and ROI from AI agents
- 20.
66% of US executives whose companies were adopting AI agents told PwC in 2025 that agents were delivering measurable value through higher productivity.
PwC, PwC's AI Agent Survey (May 2025)
Exact wording at the source
And of those adopting AI agents, two-thirds (66%) say that they’re delivering measurable value through increased productivity.
- 21.
57% of PwC respondents adopting AI agents in 2025 reported cost savings, 55% faster decision-making and 54% improved customer experience.
PwC, PwC's AI Agent Survey (May 2025)
Exact wording at the source
Over half (57%) report cost savings, faster decision-making (55%) or improved customer experience (54%).
- 22.
88% of agentic AI early adopters in Google Cloud's 2025 study saw ROI from generative AI on at least one use case, compared with 74% across all organizations.
Google Cloud, Google Cloud Study Reveals 52% of Executives Say Their Organizations Have Deployed AI Agents (Sep 2025)
Exact wording at the source
88% of these leaders report their organizations are seeing ROI from generative AI on at least one use case, compared to a 74% average across all organizations.
- 23.
69% of executives surveyed by the IBM Institute for Business Value in 2025 named improved decision-making as the top benefit of agentic AI systems.
IBM Institute for Business Value, IBM Study: Businesses View AI Agents as Essential, Not Just Experimental (Jun 2025)
Exact wording at the source
Over two-thirds (69%) of executives surveyed say 'improved decision-making' is the number one benefit of agentic AI systems.
- 24.
70% of executives in IBM's 2025 AI Projects to Profits study said agentic AI is important to their organization's future.
IBM Institute for Business Value, IBM Study: Businesses View AI Agents as Essential, Not Just Experimental (Jun 2025)
Exact wording at the source
With 70% of surveyed executives indicating that agentic AI is important to their organization's future, the research suggests that many organizations are actively encouraging experimentation.
- 25.
Capgemini Research Institute estimates AI agents could deliver up to $450 billion in economic value by 2028 through revenue gains and cost savings.
Capgemini Research Institute, Rise of agentic AI: How trust is the key to human-AI collaboration (Jul 2025)
Exact wording at the source
AI agents are poised to deliver up to $450 billion in economic value by 2028 through revenue gains and cost savings, yet the path to scale is currently elusive
- 26.
77% of business use of Anthropic's first-party API in 2025 showed automation patterns such as full task delegation, compared with about 50% for Claude.ai users.
Anthropic, Anthropic Economic Index report: Uneven geographic and enterprise AI adoption (Sep 2025)
Exact wording at the source
1P API usage is automation dominant: 77% of business uses involve automation usage patterns, compared to about 50% for Claude.ai users.
- 27.
Directive conversations, where users delegate a complete task to Claude, rose from 27% to 39% of Claude.ai conversations in Anthropic's 2025 data.
Anthropic, Anthropic Economic Index report: Uneven geographic and enterprise AI adoption (Sep 2025)
Exact wording at the source
“Directive” conversations, where users delegate complete tasks to Claude, jumped from 27% to 39%.
Chart
Top benefits of agentic AI cited by executives
Share of executives surveyed by the IBM Institute for Business Value with Oxford Economics, 2025.
- Improved decision-making69%
- Cost reduction through automation67%
- Competitive advantage47%
- Scaled employee experience44%
Source: IBM Institute for Business Value, IBM Study: Businesses View AI Agents as Essential, Not Just Experimental
Task completion benchmarks
- 28.
AI agents rose from 12% to about 66% task success on OSWorld, a benchmark of real computer tasks, but still fail roughly 1 in 3 attempts, according to the Stanford AI Index 2026.
Stanford HAI, The 2026 AI Index Report (Apr 2026)
Exact wording at the source
AI agents made a leap from 12% to ~66% task success on OSWorld, which tests agents on real computer tasks across operating systems, though they still fail roughly 1 in 3 attempts on structured benchmarks.
- 29.
Performance on the SWE-bench Verified coding benchmark rose from 60% to near 100% in a single year, according to the Stanford AI Index 2026.
Stanford HAI, The 2026 AI Index Report (Apr 2026)
Exact wording at the source
On a key coding benchmark—SWE-bench Verified—performance rose from 60% to near 100% in a single year.
- 30.
The length of software tasks frontier AI agents can complete with 50% reliability has doubled about every 7 months over six years, METR found in 2025.
METR, Measuring AI Ability to Complete Long Software Tasks (Mar 2025)
Exact wording at the source
We show that this metric has been consistently exponentially increasing over the past 6 years, with a doubling time of around 7 months.
- 31.
In METR's 2025 randomized trial, experienced open-source developers took 19% longer to complete issues when allowed to use early-2025 AI tools.
METR, Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity (Jul 2025)
Exact wording at the source
Surprisingly, we find that when developers use AI tools, they take 19% longer than without—AI makes them slower.
- 32.
The best agent tested on TheAgentCompany, a simulated software company benchmark from Carnegie Mellon University researchers, completed 30% of tasks autonomously.
Carnegie Mellon University, TheAgentCompany: Benchmarking LLM Agents on Consequential Real World Tasks (Dec 2024)
Exact wording at the source
We test baseline agents powered by both closed API-based and open-weights language models (LMs), and find that the most competitive agent can complete 30% of tasks autonomously.
- 33.
Leading LLM agents reached about 58% single-turn success on Salesforce's CRMArena-Pro business benchmark in 2025, falling to about 35% in multi-turn settings.
Salesforce AI Research, CRMArena-Pro: Holistic Assessment of LLM Agents Across Diverse Business Scenarios and Interactions (May 2025)
Exact wording at the source
Experiments reveal leading LLM agents achieve only around 58% single-turn success on CRMArena-Pro, with performance dropping significantly to approximately 35% in multi-turn settings.
- 34.
On Sierra's tau-bench customer service benchmark, even state-of-the-art function calling agents such as GPT-4o succeeded on under 50% of tasks.
Sierra, τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains (Jun 2024)
Exact wording at the source
Our experiments show that even state-of-the-art function calling agents (like gpt-4o) succeed on <50% of the tasks, and are quite inconsistent (pass^8 <25% in retail).
Chart
Leading LLM agent success rates on CRMArena-Pro
Approximate success rates of the best tested agents on a CRM business benchmark, 2025. Workflow execution figure is a lower bound.
- Workflow execution, single-turn83%
- All tasks, single-turn58%
- All tasks, multi-turn35%
Source: Salesforce AI Research, CRMArena-Pro: Holistic Assessment of LLM Agents Across Diverse Business Scenarios and Interactions
Trust, risk and governance
- 35.
Trust in fully autonomous AI agents among large organizations fell from 43% to 27% in a year, Capgemini Research Institute found in 2025.
Capgemini Research Institute, Rise of agentic AI: How trust is the key to human-AI collaboration (Jul 2025)
Exact wording at the source
Trust in fully autonomous AI agents has dropped sharply, from 43% to 27% in the past year alone, with nearly two in five executives believing that the risks of implementing AI agents outweigh the benefits.
- 36.
80% of organizations in Capgemini's 2025 agentic AI study lacked mature AI infrastructure, and fewer than one in five reported high data readiness.
Capgemini Research Institute, Rise of agentic AI: How trust is the key to human-AI collaboration (Jul 2025)
Exact wording at the source
80% lack mature AI infrastructure and fewer than one in five report high levels of data-readiness.
- 37.
74% of IT application leaders surveyed by Gartner in 2025 believed AI agents are a new attack vector into their organization, and only 13% strongly agreed they had the right governance in place.
Gartner, Gartner Survey Finds Just 15% of IT Application Leaders Are Considering, Piloting, or Deploying Fully Autonomous AI Agents (Sep 2025)
Exact wording at the source
Meanwhile, 74% believe that AI agents represent a new attack vector into their organization and only 13% strongly agreed that they had the right governance structures in place to manage them.
- 38.
Only 19% of IT application leaders in Gartner's 2025 survey had high or complete trust in their vendor's ability to protect against hallucinations.
Gartner, Gartner Survey Finds Just 15% of IT Application Leaders Are Considering, Piloting, or Deploying Fully Autonomous AI Agents (Sep 2025)
Exact wording at the source
Only 19% of respondents had high or complete trust in their vendor’s abilities to provide adequate hallucination protection.
- 39.
Only 21% of companies planning to deploy agentic AI reported a mature model for agent governance in Deloitte's 2026 State of AI report.
Deloitte, State of AI in the Enterprise 2026 (Jan 2026)
Exact wording at the source
Yet only 21% of those companies report having a mature model for agent governance.
- 40.
49% of organizations in KPMG's Q3 2026 survey had defined high-risk scenarios where autonomous AI decision-making is prohibited.
KPMG, KPMG AI Quarterly Pulse Survey (Oct 2026)
Exact wording at the source
49% have defined high-risk scenarios where autonomous AI decision-making is prohibited.
- 41.
Only 20% of US executives in PwC's 2025 survey trusted AI agents to handle financial transactions, compared with 38% for data analysis.
PwC, PwC's AI Agent Survey (May 2025)
Exact wording at the source
But trust dropped sharply for higher-stakes activities like financial transactions (20%) and autonomous employee interactions (22%).
- 42.
32% of agent builders in LangChain's 2025 survey cited quality as a top barrier to production.
LangChain, State of Agent Engineering (Dec 2025)
Exact wording at the source
Quality is the production killer, with 32% citing it as a top barrier.
- 43.
Nearly 89% of LangChain survey respondents had implemented observability for their agents in 2025, while 52% had adopted evaluations.
LangChain, State of Agent Engineering (Dec 2025)
Exact wording at the source
Nearly 89% of respondents have implemented observability for their agents, outpacing evals adoption at 52%.
Chart
Executives who trust AI agents to handle each task
Share of 308 US business executives surveyed by PwC, April 2025.
- Data analysis38%
- Performance improvement35%
- Collaboration with human team members31%
- Autonomous employee interactions22%
- Financial transactions20%
Source: PwC, PwC's AI Agent Survey
Failure, cancellation and adoption forecasts
- 44.
Gartner predicts over 40% of agentic AI projects will be canceled by the end of 2027 because of escalating costs, unclear business value or inadequate risk controls.
Gartner, Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027 (Jun 2025)
Exact wording at the source
Over 40% of agentic AI projects will be canceled by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls, according to Gartner, Inc.
- 45.
Gartner estimates only about 130 of the thousands of vendors marketing agentic AI offer real agentic products, the rest being rebranded assistants, RPA or chatbots.
Gartner, Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027 (Jun 2025)
Exact wording at the source
Gartner estimates only about 130 of the thousands of agentic AI vendors are real.
- 46.
Gartner predicts at least 15% of day-to-day work decisions will be made autonomously through agentic AI by 2028, up from 0% in 2024.
Gartner, Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027 (Jun 2025)
Exact wording at the source
Gartner predicts at least 15% of day-to-day work decisions will be made autonomously through agentic AI by 2028, up from 0% in 2024.
- 47.
Gartner predicts up to 40% of enterprise applications will include integrated task-specific AI agents by 2026, up from less than 5% in 2025.
Gartner, Gartner Predicts 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026 (Aug 2025)
Exact wording at the source
Gartner predicts up to 40% of enterprise applications will include integrated task-specific agents by 2026, up from less than 5% today.
- 48.
Gartner predicts that by 2029 agentic AI will autonomously resolve 80% of common customer service issues without human intervention, cutting operational costs by 30%.
Gartner, Gartner Predicts Agentic AI Will Autonomously Resolve 80% of Common Customer Service Issues Without Human Intervention by 2029 (Mar 2025)
Exact wording at the source
By 2029, agentic AI will autonomously resolve 80% of common customer service issues without human intervention, leading to a 30% reduction in operational costs, according to Gartner, Inc.
- 49.
Only 15% of business processes at large organizations were expected to run at semi-autonomous to fully autonomous levels within a year, rising to 25% by 2028, Capgemini found in 2025.
Capgemini Research Institute, Rise of agentic AI: How trust is the key to human-AI collaboration (Jul 2025)
Exact wording at the source
However, most deployments remain at early stages of autonomy with only 15% of all business processes operating at semi-autonomous to fully autonomous levels in a year. While this is expected to rise to 25% by 2028, most agents today function as assistants or copilots, supporting routine tasks rather than independently managing complex workflows.
- 50.
Executives in IBM's 2025 study expected AI-enabled workflows to grow from 3% of workflows to 25% by the end of 2025.
IBM Institute for Business Value, IBM Study: Businesses View AI Agents as Essential, Not Just Experimental (Jun 2025)
Exact wording at the source
The AI Projects to Profits study, which surveyed 2,900 executives globally, revealed that respondents expect AI-enabled workflows to grow from 3% today to 25% by the end of 2025.
- 51.
83% of executives in IBM's 2025 study expected AI agents to improve process efficiency and output by 2026.
IBM Institute for Business Value, IBM Study: Businesses View AI Agents as Essential, Not Just Experimental (Jun 2025)
Exact wording at the source
83% of respondents say they expect AI agents to improve process efficiency and output by 2026, and 71% believe agents will autonomously adapt to changing workflows.
Questions readers ask
- How many companies use AI agents?
- It depends on the definition. KPMG's Q3 2026 survey found 62% of large US organizations building, testing or deploying agents, and Google Cloud found 52% of executives at firms using generative AI actively using agents in 2025. Only 2% of large organizations had fully scaled agents in Capgemini Research Institute's 2025 study.
- What share of AI agent projects fail?
- Gartner predicts over 40% of agentic AI projects will be canceled by the end of 2027 because of rising costs, unclear value or weak risk controls. Deloitte found only 25% of companies had moved 40% or more of their AI pilots into production in 2025.
- How well do AI agents complete tasks?
- The Stanford AI Index 2026 reports agents reaching about 66% success on the OSWorld computer task benchmark, up from 12%. On realistic business benchmarks results are lower: about 30% of tasks on Carnegie Mellon's TheAgentCompany and about 35% in multi-turn settings on Salesforce's CRMArena-Pro.
- How much are companies spending on AI agents?
- KPMG respondents projected average AI spending of $207 million over the next 12 months in early 2026, and 88% of PwC respondents planned to raise AI budgets because of agentic AI. Google Cloud found 13% of executives at early adopter firms dedicating at least half of future AI budgets to agents.
- Do businesses trust AI agents?
- Trust is falling for full autonomy. Capgemini Research Institute found trust in fully autonomous agents dropped from 43% to 27% in a year, and Gartner found 74% of IT application leaders see agents as a new attack vector. PwC found trust lowest for financial transactions, at 20%.
Where these numbers come from
We selected figures from the organizations that ran each survey, benchmark or forecast: analyst firms, consultancies, technology companies publishing their own research, and academic or nonprofit research groups. Each figure was checked against the source page on the date shown. Samples differ a lot: KPMG, PwC and Microsoft survey executives at large companies, Capgemini and Deloitte survey global enterprises, LangChain surveys practitioners already building agents, and Gartner polls include webinar audiences. Definitions of an AI agent also vary, from agentic features inside existing software to fully autonomous systems, so adoption rates from different publishers should not be compared directly. Benchmark scores describe test conditions and the best systems available at publication, and forecasts are analyst judgments, not measurements. We store the sentence each number appears in, and a script loads the publisher's page again to make sure that sentence is still there. Our methodology has the details.
Sources
- Anthropic. Anthropic Economic Index report: Uneven geographic and enterprise AI adoption (Sep 2025).
- Capgemini Research Institute. Rise of agentic AI: How trust is the key to human-AI collaboration (Jul 2025).
- Carnegie Mellon University. TheAgentCompany: Benchmarking LLM Agents on Consequential Real World Tasks (Dec 2024).
- Deloitte. State of AI in the Enterprise 2026 (Jan 2026).
- Gartner. Gartner Survey Finds Just 15% of IT Application Leaders Are Considering, Piloting, or Deploying Fully Autonomous AI Agents (Sep 2025).
- Gartner. Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027 (Jun 2025).
- Gartner. Gartner Predicts 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026 (Aug 2025).
- Gartner. Gartner Predicts Agentic AI Will Autonomously Resolve 80% of Common Customer Service Issues Without Human Intervention by 2029 (Mar 2025).
- Google Cloud. Google Cloud Study Reveals 52% of Executives Say Their Organizations Have Deployed AI Agents (Sep 2025).
- IBM Institute for Business Value. IBM Study: Businesses View AI Agents as Essential, Not Just Experimental (Jun 2025).
- KPMG. KPMG AI Quarterly Pulse Survey (Oct 2026).
- LangChain. State of Agent Engineering (Dec 2025).
- METR. Measuring AI Ability to Complete Long Software Tasks (Mar 2025).
- METR. Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity (Jul 2025).
- Microsoft. 2025: The Year the Frontier Firm Is Born (Work Trend Index) (Apr 2025).
- PwC. PwC's AI Agent Survey (May 2025).
- Salesforce AI Research. CRMArena-Pro: Holistic Assessment of LLM Agents Across Diverse Business Scenarios and Interactions (May 2025).
- Sierra. τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains (Jun 2024).
- Stanford HAI. The 2026 AI Index Report (Apr 2026).
Citing this page
Link to this page or to a single numbered figure. If you reuse a number, name the organization that released it in the first place.
DBotMaker. (2026, November 6). AI Agent Statistics 2026. https://dbotmaker.io/ai-agent-statistics
Found a number that changed or a source that moved? Email editorial@dbotmaker.io. We plan to check this page again by January 5, 2027.