AI Agents - Value Engine

I’ve been thinking about how pricing is evolving in the context of AI agents, and I’d really value your perspective.

Historically, pricing has often been a bottleneck in the sales process—but over time, we’ve built systems (CPQ, billing, usage-based models) that make pricing more structured and scalable.

With AI agents, it feels like we’re facing a new version of that challenge—not in pricing execution, but in defining and measuring value. While costs (tokens, infra, etc.) are relatively transparent, the value created by agents is much harder to quantify consistently. We’re currently exploring ways to decode this as part of our product journey.

I’d love your thoughts on a few open questions:

Defining value: What key parameters or signals should be captured to quantify the value created by an AI agent (e.g., time saved, revenue impact, quality improvements)?
Data accessibility: In your experience, how willing are customers to share internal metrics required to measure value? What has worked (or failed) in getting this data?
Human-in-the-loop (HITL): How should we think about attributing value when humans are partially involved in the workflow? This seems like one of the hardest aspects to standardize.
Customer success criteria: How do you recommend aligning pricing with what “success” actually means to the customer, given that this can vary widely?
Product vs. sales responsibility: Should value measurement and tracking be embedded within the agent/product itself, or handled externally in the sales/pricing layer (e.g., CPQ, analytics tools)?

If you’ve seen strong frameworks, patterns, or even failed approaches in this space, I’d be very interested to learn from them.

Thanks in advance for your time—I’d really appreciate any guidance you can share.

3 comments