G2报告：AI Agent定价转向按量计费，续约前需补齐成本管控与生态能力

G2 surveyed 1,019 B2B decision-makers in July 2026 for its 2026 AI Agents Insight Report. Most of the report covers trust, oversight, and governance. The section founders should act on is chapter four, on pricing and renewals.

86% of B2B software buyers say their company bought AI agent software or services in the past year, up from 69% in 2025. 66% have agents in limited or scaled production. At roughly half of companies in most industries, agent budgets are now over $1M a year.

Only 9% of Buyers Say Per-Seat Is the Fairest Way to Price Agents

9% of buyers consider per-user or per-seat pricing the fairest model for agent services. Every other option ranked higher.

Buyer preferences shifted a lot in one year:

• Pay-as-you-go, including consumption and per completed action or task, went from 38% in 2025 to 57% in 2026
• Pay-per-outcome went from 11% to 19%
• Fixed pricing fell from 35% to 21%

A seat counts the humans who log in. An agent runs continuously and never logs in, so a per-seat price has little to do with the work it does.

Plenty of B2B companies added an agent to the product in 2025 and kept it inside the existing per-seat plan, or added a per-seat AI SKU. That was a reasonable first step. Going into 2027 renewals, it leaves you on the pricing model buyers rank last.

74% of Buyers Are Open to Replacing You at Renewal

• 30% of buyers would consider switching to a vendor with stronger agent capabilities
• Another 44% would pilot an alternative and reconsider at renewal
• 84% expect agent capabilities to influence which vendors they pick in 2027, and 34% call it a primary deciding factor

Three out of four customers are willing to run a competitive pilot against you when the contract comes up. Most B2B companies staff renewals with account managers. In 2027, plan to staff the big ones like new competitive deals.

Buyers are specific about what stronger agent capabilities means. 72% say a strong prebuilt ecosystem of connectors, actions, skills, or marketplace offerings is very important or a must-have when choosing an agent platform. They want agents, yours and other vendors’, to be able to do work inside your product.

Use of agent-builder platforms from Salesforce, ServiceNow, UiPath, Microsoft, and others rose from 38% to 42%. Use of ready-made, purpose-built agents fell from 30% to 27%. The gap between the two roughly doubled, from about 8 points to 15. More buyers are building their own agents on a platform and having those agents call into your app. If your APIs, actions, and MCP coverage are thin, those agents can’t use your product.

46% Say Agents Cost More to Run Than They Expected

Buyers want metered pricing, and almost half are already unhappy with what the meter shows.

46% of agent buyers report higher-than-expected operating costs. Agents use tokens, model calls, API calls, and data services on their own schedule, and the bill shows up afterward.

Buyers listed the controls they want before they’ll accept variable spend, in this order:

• Real-time usage dashboards
• Monthly spend caps
• Cost-per-outcome reporting
• Routing to lower-cost models
• Forecasting before deployment

Customers are fine paying for work done. They will not accept a bill that tripled because an agent looped all weekend. Launch metered pricing without hard caps and a usage dashboard, and procurement will push you back to fixed pricing.

Oversight adds to the cost. 31% of organizations say employees spend six or more hours a week supervising agents, rising to 40% at large enterprises. 45% say oversight and rework eat more than a quarter of the productivity gains they expected. A buyer paying per outcome who also spends a day a week checking those outcomes will negotiate hard on the price per outcome.

Agent Budgets Passed $1M at Roughly Half of Companies in Five of Six Industries

Share of companies budgeting at least $1M a year for agents, 2025 to 2026:

• Healthcare: 47% to 57%
• Retail: 44% to 51%
• Financial services: 42% to 49%
• Technology: 41% to 48%
• Services: 40% to 47%
• Manufacturing: 39% to 34%

I won’t invest in an agentic B2B company right now unless it unlocks budget beyond the software budget, and I want to see that in deal sizes at least 2x what the pre-AI competitors get. A company taking share of the same software line item at the same price is competing with incumbents who now ship agents too.

These numbers show the larger budget exists. Seven-figure agent line items are now common.

They also show where most of that money isn’t coming from. Only 14% of companies report a net headcount reduction from agents, and 35% say they’ve moved people to higher-value work. Most companies are funding agents as a new line item on top of existing headcount and software. When CFOs start asking for net savings, vendors with cost-per-outcome data will have an answer, and vendors reporting logins and seats won’t.

What to Change Before Your 2027 Renewal Cycle

• Plan for a competitive pilot in every top-20 account. Six months before each renewal, review measured agent results in that customer’s own workflows with them.
• Price agent work separately from seats. Keep seats for human users if you need them, and charge for agent actions, tasks, or outcomes. 57% of buyers prefer usage-based pricing. 9% prefer seats.
• Ship spend caps and a usage dashboard before metered pricing goes live. They’re the top two controls buyers asked for.
• Add cost-per-outcome reporting to the product. It’s third on the buyer list, and it’s the number a CFO will ask for at renewal.
• Put connectors, actions, and MCP at the top of the 2027 roadmap. 72% of buyers rate the ecosystem very important or a must-have, and builder platforms gained share on point solutions this year.

More Data from the Report

• In G2’s Agent Builder review data, sales development fell from 18% to 7% of use cases in a year. Process automation rose from 26% to 37%, and software engineering from 13% to 23%. It’s a review sample from one category, not market share, and it’s a bad sign for AI SDRs.
• 38% of organizations with agents in production run more than 100 of them. 47% still manage them informally or department by department, and 27% keep a central registry.
• 77% re-evaluate production agents at least monthly. A quarter do it continuously.
• Central IT plus AI and data teams now own about 60% of agent governance accountability. Vendors selling to a functional leader now also have to win IT on risk and cost control.
• 7% of buyers want agents with total autonomy. 48% want guardrails, and 29% want them limited to specific tasks.