SaaS Isn't Dead -- But Seat-Based Pricing Is_
The Apocalypse That Isn't
Every few months, a new essay declares the death of SaaS. The arguments follow a familiar pattern: AI can now do what SaaS tools do. Users will talk to agents instead of clicking through interfaces. The $300 billion SaaS market will collapse as AI replaces every dashboard, every workflow tool, every collaboration platform.
The conclusion is wrong, but the observation is right.
Something is dying. It's just not SaaS. What's dying is the per-seat pricing model — the assumption that software value is measured by how many humans log in and click around. This pricing model, which has defined SaaS economics since Salesforce pioneered it in 1999, is encountering a user that doesn't log in, doesn't click, and doesn't need a seat: the AI agent.
SaaS as a delivery model — software hosted in the cloud, continuously updated, accessible via browser and API, centrally managed — is more relevant than ever. AI agents need cloud-hosted systems to interact with. Data needs to live somewhere accessible. Workflows need execution environments. APIs need endpoints. The infrastructure of SaaS is essential to the agentic future.
But the pricing of SaaS — per seat, per month, with tiers based on features a human might use — is structurally mismatched with a world where machines are the primary consumers.
Software Becomes Labour
a16z's Big Ideas 2026 introduced the formulation that clarifies what's happening: software is eating labor. In the SaaS era, software was a tool that made human labor more productive. In the AI era, software replaces human labor with machine output.
The economic implication for pricing is direct. When software is a tool, you price it by how many humans use the tool. Per-seat pricing makes sense. More seats means more humans getting value, which means more willingness to pay.
When software replaces labor, per-seat pricing breaks. If an AI agent handles customer support that used to require 50 human agents, the company doesn't need 50 seats in the customer support platform. It might need zero seats — but it needs the underlying system (ticket routing, knowledge base, escalation logic, analytics) more than ever.
The value hasn't disappeared. It's shifted from the interface layer to the execution layer. And the pricing model needs to follow.
The Two-Week Test
Anthropic provided a vivid illustration of what happens to UI-layer tools when AI reaches sufficient capability. Their team built Cowork, an internal collaboration tool, in under two weeks. Not a prototype. Not a mockup. A functional internal tool that their team uses daily.
This isn't remarkable because Anthropic has exceptional engineers (though they do). It's remarkable because it demonstrates that the UI layer of software is approaching commodity status. When a capable team can replicate the interface and basic functionality of a collaboration tool in two weeks, the value of that interface converges toward zero.
What can't be replicated in two weeks is the data model underneath, the integrations with other systems, the compliance certifications, the years of edge case handling, the enterprise security audit trail, and the customer relationships. These are the durable assets. The UI is the disposable wrapper.
Four Durable Monetization Models
As seat-based pricing erodes, four alternative models are emerging — each aligned with where value actually lives in an agentic world:
1. Outcome-Based Contracts
Instead of charging for access to a tool, charge for the result the tool produces. A recruiting platform doesn't charge per recruiter seat — it charges per successful hire. A fraud detection system doesn't charge per analyst — it charges per dollar of fraud prevented. A code quality platform doesn't charge per developer — it charges per vulnerability caught before production.
Outcome-based pricing aligns vendor and customer incentives completely. The vendor only earns when the customer gets value. This requires confidence in delivery capability and robust measurement of outcomes — which is why it's a premium pricing model, not a commodity one.
McKinsey's research revealed a striking finding: 90% of customers chose activity-based pricing when given the choice between activity-based and traditional seat-based models. The demand exists. The challenge is defining success metrics clearly enough to build a pricing model around them.
2. Usage-Based Machine Pricing
When agents are the primary consumers, pricing calibrated to machine usage makes sense. API calls processed. Compute resources consumed. Transactions executed. Data volume accessed. Storage utilized.
This model is already proven at infrastructure companies — AWS, Stripe, Twilio all price based on usage rather than seats. The shift is extending this model upward into application-layer software that previously charged per seat.
Monday.com's introduction of AI credits is an early example. Instead of charging per user for AI features, they allocate credits that are consumed by AI actions — regardless of whether a human or an agent initiates the action. The pricing follows the usage, not the user.
3. Trust and Governance Premiums
As agents take consequential actions — spending budgets, modifying systems, communicating with customers — the governance infrastructure around those actions becomes extraordinarily valuable. Audit trails, permission frameworks, compliance certifications, rollback capabilities, identity verification for non-human actors.
This creates a premium pricing tier that has no analog in the seat-based world. Companies will pay significantly more for systems that provide verifiable, auditable, regulatorily compliant execution environments for their agents. The trust premium could exceed the base software cost — because the cost of an ungoverned agent making an unauthorized decision dwarfs the cost of the software itself.
4. Data and Access Rights
In many SaaS categories, the most valuable asset isn't the interface or even the workflow engine — it's the data and the integrations. A CRM's value isn't the dashboard; it's the customer data and the connections to email, calendar, marketing automation, and billing systems. An ERP's value isn't the forms; it's the financial data and the integrations with banking, payroll, and supply chain systems.
When agents bypass the interface, the data and access become the monetizable asset. Pricing models based on data access tiers, integration depth, and API sophistication reflect where value actually lives.
SaaS Archetypes: Survival Analysis
The pattern is consistent: the closer a product's value is to "making a human task visually easier," the more pressure it faces. The closer it is to "executing actions, managing data, or enforcing governance," the more durable it is.
UI-Heavy (dashboards, design tools, project management): Very high seat dependency. Severe agent impact — agents bypass the UI entirely. Survival path: rebuild as API-first platforms; price on usage/outcomes.
API-First (Stripe, Twilio, AWS, Sendgrid): No seat dependency. Minimal agent impact — already agent-native. Continue current model; add governance features.
Compliance-Heavy (GRC, identity, audit): Medium seat dependency. Agent proliferation increases governance demand — this is an opportunity. Add agent-specific governance; price on trust premiums.
Data Platforms (CRM, ERP, analytics): Medium-high seat dependency. Mixed impact — UI depreciates, data/integrations appreciate. Shift pricing from seats to data access and API usage.
Content Platforms (CMS, LMS, media): High seat dependency. Severe impact — AI generates content; agents consume it differently. Monetize machine readership; license structured data.
Collaboration Tools (messaging, docs, video): Very high seat dependency. Severe for human-to-human; opportunity for human-to-agent coordination. Evolve into agent orchestration platforms.
Bain's Framework: Enhance vs. Replace
Bain & Company's framework for analyzing AI impact on existing businesses distinguishes between two modes:
Enhance: AI makes the existing product better. Better search within the CRM. Smarter recommendations in the e-commerce platform. Automated reports from the analytics tool. In this mode, the product retains its value proposition and adds AI as a feature. Pricing may shift (AI credits vs. seat tiers), but the business model is recognizable.
Replace: AI eliminates the need for the product's core function. If the product exists to help humans do a task that agents can now do autonomously, the product faces existential pressure. No amount of AI features bolted onto the existing product can save it — because the product's purpose is no longer necessary.
The critical question for every SaaS company is: does AI enhance what we do, or does it replace the reason we exist? The answer determines whether the company needs to evolve its pricing model or rethink its entire value proposition.
How AI Gens Prices Ventures
AI Gens doesn't build seat-based SaaS. Every venture in our portfolio is priced around outcomes and execution guarantees, not human access to interfaces.
This isn't just a philosophical position. It's a structural decision that aligns our ventures with the direction value is flowing. When we build a data engineering platform, we don't charge per data engineer who logs in. We charge based on pipeline reliability, data quality scores, and throughput. When we build a governance framework, we don't charge per compliance officer. We charge based on audit completeness, regulatory coverage, and incident response time.
This pricing approach requires something that seat-based pricing doesn't: confidence in measurable outcomes. You can't price on outcomes unless you can define, measure, and guarantee those outcomes. This is harder than counting seats — which is exactly why it creates a moat.
The companies that figure out outcome-based pricing will own the next era of enterprise software. The ones that keep counting seats will find themselves charging for something their customers no longer need: a human sitting in front of a screen.
The Transition Is the Opportunity
The shift from seat-based to outcome-based pricing isn't instant. It's a transition that will play out over years. During this transition, the companies that can offer both models — seat-based for customers not yet ready to change, outcome-based for those who are — will capture the most market share.
But the direction is clear. Seats are an artifact of a human-centric computing era. Outcomes are the currency of an agentic era. The SaaS industry isn't dying. It's molting — shedding a pricing model that no longer fits and growing into one that does.
The winners won't be the companies that defend seat-based pricing the longest. They'll be the ones that define what outcomes look like in their category — and build the measurement, governance, and delivery infrastructure to price on those outcomes with confidence.
SaaS is alive. Seat-based pricing is on life support. The question isn't whether the transition happens — it's whether your company leads it or gets dragged through it.