An agentic layer is forming across B2B software, and it changes where value gets captured. Software has long charged humans for access, most commonly through seat-based licences. AI agents now execute significant parts of that work themselves, and two questions are arriving at the same time: how to monetise the agents you build yourself, and how to manage and monetise the third-party agents your customers are already connecting into your platform.

These questions are time-sensitive. Customer workflows settle around whichever agent arrives first, and once formed, those habits stick. Meanwhile, third-party access is already a reality. Salesforce has exposed its entire platform to external agents through Headless 360, and enterprise customers across B2B software are connecting general-purpose agents into their stack. Companies that are slow to respond risk ceding the agentic layer to someone else.

B2B software companies with regulatory accountability, proprietary domain logic, and deep customer data are well placed to build this layer themselves. These are the things that make a company's own agent more reliable and more trusted than a general-purpose alternative, and customers in complex, regulated industries will turn to vendors they already trust. They’re also precisely the kind of characteristics Hg invests behind, and it's why building agentic capabilities and getting the commercial model right is a core priority across the portfolio.

The right commercial model depends on agent ownership. When a company builds and owns the agent, it controls the outcome, and can align price to the value the customer receives rather than charging for access or inputs. That's where the highest value capture sits. When a third-party agent accesses the platform, the software company isn't delivering the outcome, and the company needs to find other ways to monetise that activity.

Pricing your own agents

Where a company builds and owns the agent, the aim should be to align price as closely as possible to the outcome delivered. Per certification confirmed. Per issue resolved. Per invoice processed without human review. This is where the commercial upside is greatest, because revenue scales with the value the customer receives rather than the cost of the inputs or the number of users.

FabricAI, a Visma company in the Hg portfolio, is one of the clearest early examples. Its agentic invoice processing for accountants charges per invoice processed without human involvement. The customer pays significantly less than the manual alternative, but because the price is anchored to the value of the outcome, FabricAI captures meaningfully more revenue per customer than under its previous pricing model. Both sides benefit.

But most companies won't start here. Outcome-based pricing requires agents that can complete tasks end-to-end without a human stepping in, and the ability to measure and bill for which outcomes the agent produced. That capability (‘metering’), is something most of the industry is still building. In the meantime, a progression is developing. Companies are launching with a fixed uplift on the existing licence that includes agentic capabilities, with generous usage allowances designed to drive adoption and gather data. As agents become more capable and metering matures, pricing moves toward charging for completed agent work.

Where agents span multiple types of activity, credits are emerging as a unifying commercial metric, with higher-value tasks consuming more credits per action. Several companies across the Hg portfolio are adopting credit-based models, and the design challenge is getting the weighting right: how many credits should a routine data lookup cost versus a complex compliance action. Companies are iterating on this as they gather real usage data.

Pricing third-party access

Even companies with strong proprietary agents will face customers building their own agent workflows, system integrators building on top, and general-purpose enterprise agents (Claude, ChatGPT) connecting into their platform. This is especially true at larger enterprises. Blocking access is rarely viable: customers will route around it or favour vendors who offer it.

For some companies, this is an offensive position, not a defensive concession. Companies whose products generate data or intelligence that feeds decisions made elsewhere (a telematics provider whose data informs fleet operations, a financial data platform that feeds investment workflows) are well positioned to become the preferred input for third-party agents. Strong technical integration and metering make that position commercially valuable.

The first decision is what level of access to give the agent. Read access, where agents can simply extract information, typically expands platform usage without cannibalising it. Allowing agents to take actions inside the platform (triggering workflows, executing business logic, modifying records) is a more consequential choice, because it allows third parties to compete in the agentic layer. How far to open up, and to whom, is one of the harder strategic questions in this space.

The second decision is how to charge. Outcome-based pricing isn't available here because the software company isn't delivering the outcome, the third-party agent is. The most common approach today is tier-gating: governing what external agents can access through existing plan levels, with agents inheriting the customer's permissions. For companies without consumption metering in place, this alone captures value by driving retention and upgrades to higher tiers, but it leaves volume-based revenue on the table. Layering consumption-based pricing on top captures more, with revenue scaling as agent usage grows.

Two of the largest enterprise platforms have independently converged on the same approach. Salesforce charges per agent action through Flex Credits, whether the action comes from its own Agentforce, a customer-built agent, or a third-party tool. ServiceNow has also opened its platform to external agents and follows the same logic: the integration is free, but every action is metered and billed through the same consumption unit regardless of what triggers it. In both cases, the commercial model is indifferent to interface.

The infrastructure that makes this work

These commercial models all depend on infrastructure that much of the industry is still building. The starting point is a clean, consistent way for agents to connect to the platform and take actions within it, including the ability to identify which agent is acting and on whose behalf.

On top of that sits consumption metering: the ability to track and bill for activity on the platform. Without this, consumption-based pricing is impossible. Access controls also need to be technically enforced, not just contractually defined, so that the tier-gating strategy has a real mechanism behind it.

This is unglamorous work, but it will separate the companies that can execute these commercial models from those that can only talk about them. Across our portfolio, we're working on these foundations in parallel with the strategic and commercial decisions above.

Why this matters now

The agentic layer creates a material new commercial opportunity. Agents that complete workflows and deliver outcomes that previously required human effort create measurable, attributable value. Companies that can price for those outcomes, rather than just for access, will capture significantly more revenue per customer.

Hg portfolio companies have structural advantages that put them in a strong position. They carry regulatory accountability, have deep trusted customer relationships, are already embedded within customer processes, and have accumulated decades of domain expertise and proprietary data to build agents. But those advantages don't convert into commercial reality on their own. They need to be matched with speed on the product side and investment in the infrastructure to track, control, and bill for agentic activity. Across the Hg portfolio, we're bringing dedicated expertise and resources to help our companies develop their agents, build the right infrastructure, and start capturing value. We’ll continue that work to evolve their commercial models as the opportunity grows.