Chicago – October 14, 2026 -- Oversight and rework consume more than a quarter of expected productivity gains for 45% of B2B organizations running AI agents, according to G2's 2026 AI Agents Report, "Crossing the Agentic Chasm." The survey of 1,019 B2B decision-makers shows 86% of software buyers purchased AI agent software or services in the past year, up 25% from 69% a year earlier.
G2 calls the cost of operating agents under constrained trust the "Distrust Tax." Some 31% of buyers say employees spend at least 6 hours a week overseeing agent work, and 46% report higher-than-expected agent operating costs.
Larger enterprises carry the heaviest oversight burden
Among larger enterprises, 27% require significantly more oversight, nearly twice the 15% rate among small and medium businesses. G2 Chief Innovation Officer Tim Sanders said the bottleneck to adoption is no longer what agents can do, but whether organizations trust them to complete the work.
Agents reach production at 66% of buyers, with measurable speed gains
Some 66% of agent buyers run agents in limited or scaled production, an 18% year-over-year increase. Four in five organizations using agents in high-volume workflows report cycle-time improvements of at least 11%, and nearly half report gains of at least 25%.
Healthcare leads overall spending, with 57% of organizations spending at least $1 million annually. Technology and retail posted the biggest rise in budgets of $5 million or more.
On workforce impact, 83% of buyers say agents improve employee satisfaction by reducing repetitive work. Some 35% have redeployed employees into higher-value roles, and 14% report reducing headcount.
Governance trails fleet growth: 38% run more than 100 agents
Among organizations with agents in production, 38% operate more than 100 agents. Some 47% say ownership and controls are informal or primarily department-led, against 23% that are centrally governed.
Only 27% maintain a central AI agent registry, 18% use automated discovery across sanctioned and unsanctioned agents, and 14% have real-time observability into unusual behavior, costs or risks. Central IT and AI, data and machine learning teams manage 49% of agent operating costs and hold 60% of governance accountability.
Pay-as-you-go pricing climbs to 57% as seat-based models lose favor
Just 9% of buyers consider per-user or per-seat pricing the fairest model for agents. Support for pay-as-you-go models, including consumption and completed actions or tasks, rose from 38% in 2025 to 57% in 2026.
Some 72% say a strong ecosystem of connectors, actions, skills or marketplace offerings is very important or a must-have. By 2030, 56% of agent buyers expect agents to orchestrate work across their SaaS stack, and 84% say agent capabilities will influence 2027 software choices, with 34% calling them a primary deciding factor.
Buyers demand proof: 77% re-evaluate agents at least monthly after deployment
Half of recent agent buyers use their own data science teams to verify vendor claims. Before production, 52% require human quality review, 48% require security testing and 41% need legal, risk or compliance review.
While 67% of buyers have high trust in their organization's AI system, 48% prefer agents with guardrails and 29% want them limited to specific tasks. Only 7% are comfortable with full autonomy.
The survey was conducted in July 2026 across North America, EMEA and APAC. Respondents defined SMBs as companies with 1-249 employees and large enterprises as those with 5,000 or more.