← All insights
AI & Pricing · Interactive
How B2B SaaS companies are pricing AI in 2026
Something significant is happening to B2B SaaS pricing, and it is happening faster than most people realise. AI is forcing the most fundamental rethink of pricing models in a decade. The conventional story is that the per-seat subscription is being replaced. The reality is more nuanced: the seat is surviving as the base commercial relationship, and companies are layering consumption and outcome-based pricing on top of it to capture AI value that a flat per-seat rate cannot reach.
Krishna Chaitanya reviewed the public pricing of more than 25 B2B SaaS products that have added meaningful AI capabilities. The market has split into five distinct approaches. The explorer below lets you try the computable ones on a portfolio company's numbers: set the number of seats and how heavily the AI is used, then watch what each model would earn and where each one breaks.
Interactive · How B2B SaaS prices AI in 2026
Five ways to charge for AI. Try them on your numbers.
Illustrative model. The per-unit rates are 2026 benchmark figures from the article (for example $0.50 to $1.00 per resolved conversation, around $25 per seat per month for a flat add-on, about $0.02 per credit). The per-seat usage volumes are assumptions you set with the sliders, not vendor data. Figures under review with Krishna before publication.
The five approaches
01
Per-seat add-on, a category in decline
A flat AI premium per seat. Microsoft 365 Copilot at $30 per user per month (a promotional $18 to $21 for business), ClickUp Brain at $9 to $28, Loom adding $5 to $8. Familiar and easy to forecast, but it overcharges the seats that barely touch the AI and undercharges the heavy users. This is the model the market is moving away from fastest: Slack, Notion and Monday.com have all retired or replaced standalone AI add-ons in the past twelve months.
02
Credits and consumption, pay for what you use
AI metered against a monthly allotment of credits or tokens, with overages priced separately. GitHub Copilot moves all plans to usage-based billing in June 2026; Figma charges $0.03 per credit, Monday.com $0.01, HubSpot $10 per 1,000. Revenue tracks actual use, but it creates budget predictability at the plan level and unpredictability at the feature level. At today's volumes it often earns surprisingly little.
03
Outcome-based, pay per result
Charge only when the AI completes the job. HubSpot Breeze at $0.50 per resolved conversation, Intercom Fin at $0.99, Klaviyo at $0.70, Salesforce Agentforce at $2.00 per conversation. The strongest value story and the highest ceiling, but it only works where the unit of value is unambiguous. Today that means customer-support resolution; for most AI capabilities no equally clean definition exists yet.
04
Hybrid: seat plus usage, where the market is heading
A predictable seat floor with consumption and outcome pricing layered on top. HubSpot runs the most deliberate version: seats for platform access, credits for AI compute, outcomes for autonomous agents, each layer mapped to a distinct type of AI activity. Klaviyo's $200 base plus $0.70 per conversation is a simpler form. This is the article's central finding: keep the seat as the anchor and build logically distinct, legible layers on top of it.
05
Bundled, AI included everywhere
AI folded into every plan at no separate charge. Xero, Zoom, Linear and Grammarly have made AI a hygiene factor rather than a premium feature. Simple and defensive, a deliberate retention and positioning play, but it leaves AI value uncounted on the invoice and may become structurally unsustainable as AI gets more expensive to run.
A sixth approach sits outside the calculator: tier-gating, where AI is the reason to upgrade rather than a separate line item (Slack moving AI into Business+, Notion into its Business tier, Pipedrive graduating AI across four plans). It does not produce a clean per-unit figure, so it is best read as a packaging decision rather than a pricing meter.
What the benchmark says
The seat is not being replaced. It is being extended. Only two of the 25 products have no seat-based component at all. What is changing is what gets layered on top.
Three patterns stand out. The flat per-user AI add-on is dying faster than expected, a transitional structure rather than a destination. The credits model is proliferating faster than customers are ready for it, training buyers to expect unpredictable line items against a fixed SaaS budget. And outcome-based pricing is real but constrained: it works for HubSpot, Intercom, Zendesk and Salesforce because a resolved support ticket is unambiguous, and stalls everywhere a clean output definition is missing.
The deepest pattern is about legibility, not mechanism. Complexity correlates with buyer friction. HubSpot runs three pricing models at once and stays legible because each layer maps to a distinct use case. Salesforce Agentforce offers five options, two of them mutually exclusive at the org level, and generates the most documented confusion. The difference is not the number of models. It is whether the architecture is designed for the buyer or for the vendor.
Why this matters for a PE-backed business
For any portfolio company adding AI, the question is not "should we replace seats?" but "what does our AI do that our seat price does not capture, and what is the right mechanism to capture it?" Match the layer to the capability: agentic AI that delivers measurable outcomes warrants outcome-based pricing; assistive AI that improves existing workflows suits tier inclusion; highly variable, token-heavy workloads warrant a credits layer. The failure mode is picking a mechanism because it is fashionable rather than because it fits the value delivered.
And do not price for Year 1 adoption at the expense of Year 3 monetisation. Buying usage cheaply is rational if you believe it will become habitual, but every company doing it needs a credible path to monetisation within 18 to 24 months, or the capability becomes a structural cost with no matching revenue. Inside a hold period, getting that architecture right before competitors lock in the category norm is one of the most consequential commercial decisions a SaaS business will make in 2026.
Built on the benchmark article "How B2B SaaS companies are pricing AI in 2026" by Krishna Chaitanya, independent pricing and commercial strategy consultant. Data sourced from public pricing pages, May 2026. The full written analysis is published on Hirondl Insights.
Working through AI pricing in a portfolio company?