Billing Infrastructure in the Age of Co-pilots and AI Agents

The dawn of intelligent systems, from “selling the software” to “selling the work,” is redefining economic value. These changes don’t just affect pricing—they ripple across every facet of business operations.

As we move toward co-pilots and autonomous AI agents, pricing models must evolve. We have continued to shift further away from predominately human usage, priced on a per-seat license, to a world with an increasing number of human-to-machine and/or machine-to-machine-based interactions. The rise of AI agents performing tasks autonomously will require a re-evaluation of economic value derived from underlying services.

AI software could be priced based on compute, similar to databases. It could also be priced on an outcome-based approach, with closer value alignment. This will necessitate a new billing infrastructure. The seat based model once disrupted the perpetual license model, so too will usage/performance pricing disrupt SaaS. In a decade, we may look back at the era of subscription pricing as an anomaly.

This shift to usage or performance-based pricing requires a data infrastructure-first approach to align with process flow.

Shifting to usage-based pricing means that billing can no longer be an afterthought—it must be built on a robust data infrastructure from the ground up. Every billable action (API call, computation, transaction, etc.) needs to be metered, rated, and invoiced. This requires robust handling of high volumes of event data, applying agile monetization rules, and integrating with finance systems in near-real-time. We will see the emergence of data observability for agile monetization.

There is a clear analogy with that of telco usage-based billing that had to re-design its infrastructure to accommodate large volumes of data usage, complex billing systems (e.g., tiered rates, discounts, promotional offers, bundled services), real-time processing, regulatory compliance, and reliability, etc.

Table 1. For illustrative purposes only. Dollar values are representative of global Total Addressable Market (TAM) as of 02/2025; they are estimates and subject to change. Logos are not representative of Lightspeed portfolio companies. Sources: See end of article.

Usage-based pricing’s ripple effects across the billing infrastructure stack

Monetization is now an all-encompassing process that spans the entire organization, impacting product development, marketing, sales, finance, billing, and customer support—extending from the back office to the front office and into the customer and supplier ecosystem. Indeed, usage-based pricing impacts everything between the CRM and the ERP:

CRM

Traditional CRM systems (e.g., Salesforce, Microsoft Dynamics) are designed around human sales reps managing relationships with customers. However, these interactions are no longer purely human-driven.
In the old model, a salesperson’s goal was to sign a big annual contract (e.g., $X for Y seats), and they’d earn a commission on that contract value. In a consumption model, however, value unfolds over time as the customer uses the product.
In essence, the CRM now also tracks telemetry about how customers use the product (since that ties directly to billing and renewal conversations).

CPQ

CPQ systems help sales teams generate complex quotes for configurable products. Pricing changes for each SKU product are typically hard-coded and can take 3-6 months to implement.
In this new paradigm, pricing and packaging need to be agile to support a wide array of models (per-use, tiered plans, volume discounts, overage pricing, prepaid credits, outcome-based charges, etc.) via configuration rather than code; for example, software licensing, cloud services, and enterprise solutions.

CLM

CLM systems manage legal contracts manually, requiring human review for terms, compliance, and risk assessment.
We’re seeing a shift to more autonomous contract review systems, clause extraction, and dynamic contract negotiation.
These systems will need to be better integrated with AI-driven procurement, billing, and revenue recognition systems

Procurement

We are moving away from rule-based approval workflows to intelligent systems negotiating autonomously, albeit with supervision or pre-determined utility curves.
Spend analytics, supplier risk assessment, and demand forecasting will all up-level.

Billing/Invoicing

Moving towards dynamic billing based on real-time consumption. However, this will require a greater focus on the end-customer experience of billing. Rather than a customer being surprised by a bill at month’s end, these systems enable real-time visibility of usage and spend.

Revenue Management

Traditional revenue management systems (e.g., NetSuite, Zuora, SAP BRIM) are built around predefined pricing models and predictable revenue recognition rules.
From an accounting perspective, usage-based revenue introduces new complexities. Finance teams can no longer recognize revenue evenly based on a straightforward annual subscription.
Systems will need to automate complex revenue recognition for AI-driven and usage-based billing models, and ensure compliance with ASC 606, IFRS 15, and other financial regulations in real-time.
We’ll see more autonomous financial operations, including accounting close management, business planning, revenue reconciliation, etc. Multi-dimensional modeling may finally enable revenue and cost matching at a micro-level.

ERP

The evolution in billing and monetization opens a window to rethink legacy ERP systems as well. These tools were not built with agile pricing in mind. They assumed relatively static pricing, simple contract terms, and minimal mid-stream changes. Usage-based pricing exposes their inflexibility.
We often see finance teams exporting data out of an ERP to spreadsheets for adjustments, or building ancillary databases to handle usage ratings because the core system can’t do it.
AI disrupts static planning, financial reporting, procurement, and supply chain management/inventory control by enabling real-time, autonomous decision-making.

Table 2. For illustrative purposes only. Dollar values are representative of global Total Addressable Market (TAM) sizing as of 02/2025; they are estimates and subject to change. Logos are not representative of Lightspeed portfolio companies. Sources: see end of article.

It changes revenue reconciliation and requires changing sales compensation plans, product development, servicing engineers with robust backend systems, intuitive frontend interfaces for sub-function beneficiaries, and revenue visibility for finance.

Value will be re-distributed in this new paradigm. Billing = the new revenue and cost intelligence layer.

In a usage-based model, the most accurate and up-to-date source of customer value isn’t a static contract in a CRM—it’s real-time product usage data. The billing infrastructure becomes the data observability for the broader organization.

CRM and ERP are not going away, but they will adapt.

As AI makes unstructured data (e.g., product usage logs, emails, call transcripts, and customer interactions) more accessible and actionable in real-time, the need for static, monolithic depositories—like traditional CRMs or ERPs—diminishes. Instead of storing data in rigid, predefined structures, modern systems are moving toward ephemeral, event-driven architectures. Usage-based billing means costs are also dynamic–they depend on factors like compute usage, API calls, etc. As a result, companies with flexible, AI-powered billing can experiment with new pricing strategies faster, be it tiered usage, per-event billing, or surge pricing, unlocking new economic value and agility.

The content here should not be viewed as investment advice, nor does it constitute an offer to sell, or a solicitation of an offer to buy, any securities. Certain statements herein are the opinions and beliefs of Lightspeed; other market participants could take different views.

Table Information Sources:

Writing Archive:

Authors

Adrian Radu