Recent advancements in AI are reshaping software — yet one fundamental question remains: how do we price it? The pricing landscape for AI includes a variety of approaches, lacking a unified standard. Whether you’re an old school SaaS incumbent, or an AI-first product, companies are still exploring different approaches. What’s clear is that the vanilla seat-based pricing model of the last decade is on its last legs. The dynamic nature of AI poses a challenge: Will the industry gravitate towards a single pricing strategy? And are we going to see more permutations of pricing models?

For years, companies have been looking to shift away from basic subscription pricing towards models that tie cost directly to value. We’ve seen this in APIs ($ per request, like Twilio), storage ($ per GB), compute power ($ per minute), and communication
($ per email/SMS, like MessageBird).

Yet, pure usage-based pricing hasn’t fully taken off. Why? Internally, tracking and billing require massive engineering effort. Externally, customers want predictable costs. Striking the right balance — offering value-based pricing without creating uncertainty — has been a challenge. A model needs to work for sales, customer success, engineering, and finance.

This is where outcome-based pricing comes in. Instead of charging for access or usage, companies charge for results — aligning cost with business impact.

A new generation of software companies is born

As AI agents become more capable, they are forcing a split in how SaaS companies operate and monetise their offerings. Some are layering AI onto existing products, while others are building AI-native solutions that function more like autonomous workers than traditional software. Generally, they can be divided into:

• AI-assisted services: Companies that layer AI features on top of existing software to act as a copilot for the user

• AI-native services: Companies built from the ground up with AI agents that execute work autonomously

The distinction isn’t just technical — it also changes how businesses operate commercially. AI-assisted tools can still justify per-seat pricing, but AI-native companies need a different playbook. They can’t charge per user when users no longer exist.

As a result, we currently see software companies structure their pricing in various ways.

SaaS companies are facing the AI margin trap

Traditional SaaS thrived on high margins — low variable costs, high recurring revenue. AI is flipping this model upside down. Three major forces are at play:

• Rising Compute Costs: AI-powered software features require a high level of compute resources. Each AI-generated response, workflow execution, or inference comes at additional cost.

• Inflated Customer Expectations: AI-powered features are now expected, not premium. As a result, charging higher price points becomes difficult.

• Shrinking User Base: AI automation removes humans from workflows, slashing the number of billable seats.

SaaS companies that fail to adapt to this new dynamic will see margins collapse.

Salesforce is proving that also more established SaaS companies are accepting the new reality. With Agentforce, Salesforce launched an agentic offering in September 2024. Within weeks, it secured 1,000+ paid deals, adding $4B in projected 2025 revenue.

Their pricing? $2 per agentic conversation, regardless of impact. Should higher-value outcomes cost more? Most likely. But as Intercom and Zendesk are also adopting similar models, it is at least clear that there is an ongoing shift away from seat-based pricing.

Pricing AI for outcomes, not effort

Think about customer service centers. Traditional BPOs charge per minute. But what actually matters to customers? Getting issues resolved.

Take our portfolio company Parloa, for example. Their customer service agents don’t just close tickets — they can upsell customers, and boost customer satisfaction. Their AI agents don’t just replace human labour — they unlock new revenue streams.

So how can you capture the full value of your agentic workforce, and make outcome-based pricing work in practice? Three key steps:

1. Define the right outcome. Treat AI agents as a revenue-driving workforce, not just a cost-saving tool. Measure what really matters — customer retention, sales conversions, or efficiency gains.

2. Make attribution crystal clear. Customers won’t pay unless they trust the results. Build transparent tracking, audit trails, and customer validation into your pricing.

3. Create cost predictability. Enterprise buyers need budget certainty. Tiered pricing, spend caps, and hybrid models help ease adoption.

Moving from predictable pricing to outcome-based models isn’t easy. It requires a mental shift from treating agentic AI applications as a cost item to a new revenue driver; from being a “software vendor” to a true “outcome partner” that charges based on results.

This is not just a shift to a simple metering event — it’s about redefining how value is measured. Consider complex outcomes, such as upsells, customer retention, or efficiency gains, each requiring distinct pricing structures for different customer tiers. In an outcome-based world, billing can no longer be an afterthought or a simple extension of existing models.

After many discussions with teams operating in this space, one thing has become clear: existing billing engines fall short. Capturing the true value of AI-driven outcomes requires a fundamental rethink. The next generation of billing infrastructure must be built from the ground up to be AI-native, not just AI-enabled.

We have seen first companies moving in this direction — but critical questions remain unclear:

• How do we define and price outcomes fairly?

• How do we build transparent attribution mechanisms?

• How do we set appropriate price ceilings to protect customers?

The companies that answer these questions will shape the future of software. Those that don’t risk being left behind.

As a team, we’re looking at backing the companies building the picks and shovels for the agent economy. If you are working on a related topic, feel free to reach out at doreen.huber@eqtventures.com, kaushik@eqtventures.com, or conrad.schoo@eqtventures.com.

We would love to chat.