Incumbent Agentic Pricing: How Salesforce and HubSpot Defend Per-Seat Revenue Models
As autonomous agentic AI systems become deeply integrated into enterprises in 2026, legacy SaaS incumbents are caught in a classic innovator's dilemma: if their AI agents genuinely replace human work, their customers will eventually require fewer human seats. To defend their revenues while capitalizing on AI adoption, giants like Salesforce and HubSpot are aggressively experimenting with pricing architectures, combining seat-based floors with consumption- or outcome-based metrics.
Salesforce Agentforce: Three Parallel Pricing Models
To navigate the transition, Salesforce has deployed three distinct pricing models for Agentforce simultaneously, allowing enterprise customers to self-select based on their procurement preferences:
- The $2 Per-Conversation Model (October 2024): This model was elegant in theory but failed in production. Single customer queries often triggered multiple backend processes, making costs highly unpredictable. Nonprofits and SMBs were immediately priced out, and only 3,000 of the first 5,000 Agentforce deals were paid.
- Flex Credits at $0.10 Per Action (May 2025): Salesforce pivoted to a more granular action-based credit model (e.g., 100,000 credits for $500). Each action—such as updating a record, summarizing a case, or resolving an inquiry—consumes roughly 20 credits. While more aligned with actual work, it remained consumption-based and difficult for enterprise procurement to forecast.
- Agentic Enterprise License Agreements (Late 2025): Salesforce introduced flat per-user licenses starting at $125/user/month to package AI agents as "digital labor." This model gives CFOs budgetary predictability and fits traditional enterprise procurement workflows.
This multi-model strategy has driven Agentforce to $540 million in ARR by Q3 FY2026, representing a 330% year-over-year growth. However, only about 8% of Salesforce's 150,000+ customer base has adopted the tool so far.
The Cannibalization Threat
The threat to legacy per-seat revenues is already materializing. Salesforce sales engineers across 90 enterprise accounts report an average 10% reduction in human seats/headcount because AI is making human customer service agents significantly more efficient. Salesforce itself is leading this transition: its internal deployment of Agentforce handled over 380,000 customer support interactions, fully resolving 84% of them without human intervention, and leaving only 2% for human escalation.
HubSpot Breeze AI: The $0.50 Per-Resolution Illusion
In April 2026, HubSpot shifted its Breeze Customer Agent and Prospecting Agent to an outcome-based pricing model of $0.50 per resolved conversation (50 credits), billed only when the AI successfully resolves a ticket without human intervention within 72 hours. This replaced its previous model of 100 credits (~$1.00) per conversation regardless of resolution.
While $0.50 per resolution is the cheapest headline rate among major helpdesk AI agents (compared to Intercom's $0.99 and Zendesk's $1.50+), the true cost is heavily bundled:
- Seat Floor: Running Breeze Customer Agent still requires a Service Hub Professional or Enterprise seat (starting at $90/seat/month).
- Onboarding Fees: HubSpot charges a mandatory, one-time onboarding fee of $1,500 for Pro and $3,500 for Enterprise.
- Credit Expiration: Purchased credits expire monthly with no rollover.
A customer service team handling 10,000 conversations a month at a 50% resolution rate would pay roughly $2,500 in resolution credits and $1,800 in human seats, totaling $4,300/month before onboarding. This structure ensures that HubSpot maintains a high, predictable seat-based revenue floor even as its AI agents drive down the total human headcount required by its customers.