AI agents are making enterprise software more productive than ever.
Tasks that once required multiple employees can now be completed with the help of intelligent automation.
For customers, that's a huge advantage.
But for SaaS companies using per-seat pricing, it introduces a new business challenge:
What happens when customers need fewer seats because AI is doing more of the work?
This emerging trend—often called seat compression—is changing how product teams, engineering leaders, and founders think about long-term SaaS growth.
Some of the key risks include:
• Slower seat expansion despite higher product usage
• Reduced ARR growth from enterprise accounts
• AI replacing repetitive workflows previously handled by multiple users
• Pricing models becoming misaligned with customer value
• Increased pressure to demonstrate measurable business outcomes
• Investors focusing more on NRR than license growth
• A growing shift toward usage-based and value-based pricing
One of the biggest misconceptions is that adding AI automatically increases SaaS revenue.
In reality, AI can help customers accomplish more with fewer users—which is great for them but can challenge traditional seat-based business models.
For engineering and product teams, success isn't just about building smarter AI agents.
It's about creating products that become essential to customer workflows, deliver measurable ROI, and continue generating value regardless of the number of seats purchased.
The future of SaaS may depend less on how many users a company has and more on how much value each customer receives.
I've shared a detailed guide explaining how AI agents are driving enterprise seat compression, the risks for SaaS companies, and strategies to adapt pricing and product strategy for long-term growth:
https://mavanisolution.com/resources/ai-agents-enterprise-seat-compression-risk
Question for the DEV community:
As AI agents become more capable, do you think SaaS companies should continue using per-seat pricing, or is it time to shift toward usage-based or outcome-based pricing models? Why?
Top comments (0)