New G2 Research: AI Agents Have Moved From Pilot to Production, But Oversight Comes at a Cost

• 86% of B2B software buyers purchased AI agent software or services in the past year, a 25% increase from the previous year.
• 31% say employees spend 6 or more hours each week bot-sitting (overseeing agent work), and 45% say that time consumes more than a quarter of the productivity gains they expected.
• 38% of organizations run 100+ agents, and half have an agent budget of at least $1M, yet 47% still manage them informally or through decentralized models.
• 84% of buyers say agent capabilities will influence their 2027 software choices, with 34% considering them a primary deciding factor.

CHICAGO, Oct. 14, 2026 /PRNewswire/ -- G2, the world's largest and most trusted data source on B2B software, today published its 2026 AI Agents Report, titled "Crossing the Agentic Chasm," which reveals where AI agents are delivering value, why some of those gains are eroding in production, and what businesses need to build trust and enable adoption at scale. Based on a survey of over 1,000 B2B decision makers, the report finds that 66% of B2B agent buyers now have agents in limited or scaled production, an 18% year-over-year increase. Yet, the productivity gains aren't always keeping pace, as 45% say oversight and rework consume more than a quarter of expected gains.

G2's findings from its 2025 Agents Report identified the Trust Gap, which stems from a lack of trust in agents despite their high reliability. The 2026 report reveals that this Trust Gap has led to a Distrust Tax — the cost of operating agents under constrained trust. G2's study indicates one-third of B2B organizations with agents in production spend at least 6 hours per week overseeing agent work, and 45% say oversight and rework consume more than a quarter of their expected productivity gains. For agents to reach mass market adoption, more evidence about how agents work in their systems, clear ownership, and cost controls can help lower the Distrust Tax.

"AI agents are showing real signs of value, but much of it is being erased by oversight, rework, and exception handling," said Tim Sanders, Chief Innovation Officer at G2. "An agent's ability to do work is no longer creating the bottleneck to adoption; it's now formed when organizations don't trust agents to complete the work. Absent trust, agents are limited to isolated use cases and are slowing the market's move into mainstream adoption. For agents to cross that chasm, pragmatists need real proof of credibility — not more marketing promises, but repeatable results in real workflows and the controls that make autonomy safe to extend."

Key findings from G2's 2026 AI Agents Insights Report include:

AI Agent adoption is advancing from pilots to production

Agents are moving beyond experimentation into workflows where companies can measure speed, quality, and cost.

• AI agent adoption by B2B software buyers rose 25% year over year, from 69% to 86%, and 66% of companies now run agents in limited or full production.
• Technology and retail saw the biggest rise in $5 million+ budgets, while healthcare leads the overall spending with 57% of organizations spending at least $1 million annually.
• 4 in 5 organizations using agents in high-volume workflows report at least 11% improvements in cycle time, while nearly half see gains of at least 25%.
• More than 8 in 10 (83%) buyers say agents improve employee satisfaction by reducing repetitive work, and 35% have redeployed employees into higher-value roles. Just 14% report reducing headcount.

The Trust Gap has created a "Distrust Tax"

The agent Trust Gap has begun to narrow as agents advance into production. However, bot-sitting and rework consume a significant portion of agent gains.

• Nearly one-third (31%) of B2B buyers say employees spend at least 6 hours each week overseeing agent work.
• 45% report losing more than a quarter of expected agent productivity gains due to oversight and rework, while 46% report higher-than-expected agent operating costs.
• The Distrust Tax hits larger enterprises hardest: 27% require significantly more oversight, nearly twice the SMB rate of 15%.

Shadow Agents have become the new Shadow IT

Organizations lack visibility into which agents their employees use, who owns them, and the risks they may pose. As agent fleets grow, inventory, identity, visibility, and accountability are lagging.

• 38% of organizations with agents in production operate more than 100 agents.
• Almost half (47%) say agent ownership and controls remain informal or primarily department-led, compared to just 23% that are centrally governed.
• Just 27% maintain a central AI agent registry, while 18% use automated discovery across sanctioned and unsanctioned agents, and just 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 collectively own 60% of governance accountability.

AI Agents are disrupting SaaS

Agents are changing both the products buyers want in their software stack and how they expect to pay for them. Traditional seat-based pricing models and disconnected products are becoming weaker fits for agent-driven work.

• Just 9% of buyers consider per-user or per-seat pricing the fairest model for AI agents.
• Pay-as-you-go models, including consumption and completed actions or tasks, rose from 38% in 2025 to 57% in 2026.
• 72% say a strong ecosystem of connectors, actions, skills, or marketplace offerings is very important or a must-have for agent adoption.
• By 2030, 56% of agent buyers expect agents to orchestrate work across their SaaS stack.

AI agents face a higher bar for proof than traditional software

Beyond testing, the real challenge begins when AI agents enter day-to-day workflows. Buyers need to understand what agents can reliably handle on their own and when human intervention is necessary.

• Half of recent agent buyers use their own data science teams to verify vendor claims.
• 4 in 10 ran a hands-on test or proof of concept, made reference calls, consulted analyst evaluations, or used independent benchmarks.
• Before production, 52% require human quality review, 48% require security testing, and 41% need legal, risk, or compliance review. After deployment, 77% re-evaluate agents at least monthly.
• 67% of B2B software buyers have high trust in their organization's AI system, but 48% prefer agents with guardrails, 29% want them limited to specific tasks, and only 7% are comfortable with full autonomy.

Read more findings from G2's 2026 AI Agents Report and explore G2's recently launched beta version of AI Agent Evaluations to see how G2 is building trust for the agentic economy.

About the report

G2 fielded an online survey among 1,019 B2B decision-makers responsible for or influencing purchase decisions for departments, multiple departments, operating units, or entire businesses. Respondents had job titles ranging from individual contributor to manager, director, vice president, or higher.

To maximize differentiation for vendors, this survey defines small-medium business (SMB) as a company with 1-249 employees, mid-market as a company with 250-999 employees, enterprise as a company with 1,000-4,999 employees, and large enterprise as a company with 5,000+ employees. The survey was conducted in July 2026 and included a global pool of respondents across North America, EMEA, and APAC.

This report also cites software review data from G2.com. Over 2,500 software reviews in G2's AI Agent Builder category from March-August 2026 were referenced to illustrate agent use cases. Generative AI reasoning models were used to define the study's focus areas, optimize survey design, and analyze results to inspire writing and data visualizations.

About G2

G2 is the world's largest and most trusted data source on B2B software, helping businesses reach their peak potential by enabling confident buying and go-to-market decisions. Offering trusted data, authentic peer reviews, and real-time market intelligence, the G2 ecosystem — which includes Capterra, Software Advice, and GetApp — serves more than 200 million annual buyers, representing teams at every Fortune 500 company.

As buyers increasingly shift from traditional search to AI search platforms, G2 has become the most-cited B2B software source across those AI-first channels where software discovery happens. Leading software and services companies like Salesforce, IBM, SAP, Adobe, and Clay also trust G2 to influence discovery, build brand credibility, reach in-market buyers, and accelerate revenue growth. To learn more, visit www.g2.com and follow us on LinkedIn.

View original content to download multimedia:https://www.prnewswire.com/news-releases/new-g2-research-ai-agents-have-moved-from-pilot-to-production-but-oversight-comes-at-a-cost-302908654.html

SOURCE G2