Brandon Hickie on AI Pricing, Credit Economies, and the Rise of the Orchestration Layer
Brandon Hickie leads monetization for LinkedIn’s Talent Solutions business and is also a venture partner at Companion Ventures, where he advises growth-stage companies on go-to-market strategy and commercial models. Earlier in his career, he worked in antitrust and M&A strategy at Charles River Associates and later advised expansion-stage companies through OpenView Venture Partners.
In this conversation, Hickie explains why generative AI and autonomous agents are forcing companies to rethink pricing, packaging, governance, sales incentives, and commercial infrastructure. His central argument is that the agent economy will not be defined by a single pricing model. Instead, it will be shaped by hybrid monetization, simpler and more flexible credit systems, and orchestration layers that unify how customers buy, govern, and deploy AI capabilities.
Disclaimer: The views expressed here are Brandon Hickie’s personal views and do not represent the views of LinkedIn, Companion Ventures, or any other organization. Examples are illustrative only.
Key points
AI is changing the economics of scale. The relevant unit of productive capacity is no longer simply the employee or software seat; it is the employee plus the AI tools and agents that amplify their output. This forces CFOs to assess AI spending against the cost and value of human-led alternatives.
Hybrid monetization will be the practical default for most established software companies. Rather than moving immediately from seats to pure consumption or outcomes, many companies will combine subscriptions, seats, credits, usage charges, and selectively priced AI capabilities.
Usage pricing is an operating model, not just a billing model. As AI consumption becomes a meaningful share of customer spend, vendors need near-real-time usage data, spend controls, administrative tools, self-service replenishment, and sales incentives that support consumption-based growth.
Credits will persist—but only if they become much simpler. Customers will reject an accumulation of incompatible credit pools, opaque conversion rules, and unpredictable costs. The likely direction is toward common credits that can be used across multiple workflows, supported by a sophisticated back-end value and margin model.
Autonomous agents make governance part of the product and commercial model. Enterprises will require clear controls over what agents can access, spend, and do—especially in regulated environments or when agents handle sensitive data.
The biggest strategic opportunity may be the orchestration layer. As systems of record open to third-party agents and services, the platform that unifies interaction, purchasing, governance, and credit management could capture the deepest moat and become the economic control point for its ecosystem.
From pricing to monetization
My path into pricing began indirectly, through antitrust and M&A strategy. I was working on questions of market definition, market evolution, and the risks created when companies acquire to protect market share. That required thinking about how innovation would move, how markets might develop, and what competitive structures might emerge over time.
That perspective has carried through my work in pricing and monetization. At LinkedIn, I have seen the function evolve from a small internal pricing consultancy into a strategic monetization organization. Through Companion Ventures, I have also had the opportunity to advise companies building in AI infrastructure and adjacent categories.
The common thread is that pricing is never only about the price itself. It is about market structure, incentives, value creation, and the choices a company makes about where it wants to compete.
AI changes the economics of scale
We are still in the early innings of AI. In many cases, AI is being added to existing workflows to improve efficiency rather than to create entirely new sources of value. Companies are bolting AI onto products, trying to help users complete work faster, reduce cost, or improve output.
A smaller set of companies has moved further. They have figured out how AI can unlock markets or business models that would otherwise have been unavailable to them. Those are the true leaders, because they are not merely using AI to make an existing process more efficient. They are using it to create new value.
This has major implications for monetization. Seat-based pricing assumes a relatively stable relationship between the number of people using a product and the value it produces. AI breaks that assumption. Value is not distributed evenly. A small number of power users—or a small number of automated workflows—can create dramatically more value than everyone else.
“Scale and growth are no longer just about human capital. They are about human capital plus AI.”
That changes how a CFO evaluates return on investment. The relevant comparison is increasingly not just software cost versus software benefit. It is the cost of doing work through people versus the cost of doing work through people augmented by AI—or, eventually, through AI acting with increasing autonomy.
As token and model costs become more visible, buyers will become more disciplined. CFOs will ask whether the value created justifies the cost. If it does not, they will remove tools. That will lead to consolidation, especially where customers face complex buying, user, or management processes.
Why hybrid models will prevail
One of the clearest trends is the move toward consumption. AI-native companies without large installed revenue bases can often begin with a consumption model. That is much harder for established companies. Moving an existing customer base from a stable subscription model to variable consumption can create disruption, uncertainty, and large swings in customer cost.
That is why hybrid models are likely to become more common. Rather than moving entirely to consumption or entirely to outcome-based pricing, established companies will combine a core subscription or seat-based model with usage, credits, premium AI capabilities, or selected outcome-linked elements.
A hybrid model gives an incumbent room to create new value without forcing a complete commercial migration all at once. It can also help preserve a predictable revenue base while allowing the company to learn where consumption pricing works and where it does not.
The key question is what kind of value an AI capability creates.
If it primarily drives efficiency and is likely to become a table-stakes capability, bundling it into the core product may make sense.
If it creates distinct, measurable, and incremental value, there may be a stronger case for a separate usage or value metric.
If its economics are predictable and usage will grow consistently, a consumption component may be appropriate.
Not all AI workflows should be priced in the same way. Some create net-new value. Some simply reduce effort. Some have predictable economics that will grow steadily with usage. The commercial model needs to reflect those differences.
For many companies, AI will initially function as the scaffolding around the house while they rebuild the underlying product and business model. In those cases, bundling may be the correct strategic choice.
Consumption requires control
The most important change in AI monetization is the need for control. In an enterprise, the buyer is often not the user. That creates a disconnect: the person who experiences the value is not necessarily the person accountable for the purchase.
With AI, that becomes more acute. If consumption is pooled across a company, a small number of users can consume a disproportionate amount of capacity. In a traditional seat-based model, an organization can simply decide not to provision a seat. In a usage-based model, it needs much better ways to manage what is being consumed, by whom, for what purpose, and within what limits.
That is why administrative controls, spend visibility, budgeting, usage limits, and approval workflows will become increasingly important. Companies will need customer-facing dashboards that are close to real time. They will need to give finance leaders and administrators the ability to control usage rights and allocations or a mechanism to monitor see consumption, assess value, and intervene before a low-value workflow becomes a large cost problem.
When consumption becomes material—when it represents 20 percent or more of a customer’s cost structure—the operating requirements rise sharply. What may have been acceptable when AI represented a small add-on becomes inadequate when it becomes a meaningful portion of spend.
This is not simply a billing problem. It affects revenue operations, finance, procurement, customer success, and sales compensation.
Salespeople will naturally prefer a predictable subscription model if their compensation is tied to contracted revenue. If a company wants its sales team to sell consumption, it needs to redesign incentives so that the team is not taking more risk for less predictable compensation.
A company also needs to be realistic about ROI measurement. AI supports an enormous number of workflows, and a single universal ROI model is unlikely to work. The work is to define the use cases that matter, determine how value will be measured, and connect spending to the business outcomes that customers actually care about.
Autonomous agents raise the stakes
Autonomous agents increase the importance of governance because they can take actions, consume resources, and interact with data without a person actively supervising every step.
Organizations will need to understand what agents are permitted to consume, what they are allowed to do, where they can operate, and which data they can access. This will be especially important in regulated environments and wherever personal, financial, health, or other sensitive data is involved.
Companies that can establish clear policy controls around agents and create appropriately governed environments will have an advantage. They will be better positioned to satisfy enterprise security requirements and operate in regulated markets.
The central challenge is to create rules that allow organizations to capture value while managing risk. That applies both to enterprise governance and to public policy. As AI creates larger disparities among winners and losers, pressure for regulation will increase. The companies and platforms that help define workable governance models will be in a stronger position than those that treat regulation as an afterthought.
Credits are here to stay
Credits are likely to remain part of the commercial landscape for the next decade. Not every company will be able to attribute value cleanly enough to charge on outcomes. The cost of defining, measuring, and debating an outcome can be prohibitive—particularly for mid-market companies or lower-priced products.
Credits provide a workable path between a fixed subscription and a fully outcome-based model. But there are different kinds of credits, and that distinction matters.
One model treats credits as prepaid capacity. The customer purchases a pool and can apply it across defined services. Another treats credits as a more transparent proxy for dollars spent. These models create different expectations for finance teams. The closer credits are to a direct dollar wallet, the more governance and control buyers are likely to demand.
“Credits are here to stay—but only if companies make them simpler.”
The success of credits will depend on simplicity. The market is at risk of creating death by a million credits. If every product has its own credit type, rules, conversion rates, and expiration policy, customers will face the same frustrations they experienced with unused software seats—except now the complexity is even greater.
A customer should not need to become an expert in a vendor’s internal cost structure simply to understand what they are buying.
The case for common credits
A better credit system is likely to be built around common credits: a single unit that can be applied across different workflows, even though those workflows consume different resources and create different kinds of value.
For example, one credit might be used for an automated email, a set of AI-agent searches, or part of an orchestration workflow. The user sees a simple, common unit. Behind the scenes, the provider maintains a value table that determines how credits translate across services with different cost and margin structures.
This is easier for the customer, but it creates a more difficult internal economics problem for the vendor. If credits are interchangeable across products, business lines, or an entire portfolio, the company needs sophisticated instrumentation to allocate revenue, control margins, and understand where value and cost are moving.
The principle is simple: the customer will only adopt the system if it is simple enough. The back-end economics can be complex, but the front-end experience cannot be.
Companies will also need to provide real-time dashboards, administrative controls, and self-service purchasing. Credits are often purchased in smaller increments and on drawdown schedules rather than through a conventional monthly recurring-revenue model. That changes revenue recognition, purchasing behavior, and the way finance creates buying authority.
From credits to credit ecosystems
The most important development may be the dollarization of credits: the emergence of a platform that becomes the default credit system for an entire functional ecosystem.
Think about how a shared currency simplifies exchange across a region. In the agent economy, a common credit system could simplify purchasing, management, and interoperability across many applications and services. A platform could act as a credit broker, enabling customers to buy once, manage one pool of value, and allocate it across multiple vendors and workflows.
The analogy is the evolution of platform ecosystems such as app marketplaces. A platform does not need to build every capability itself. Its value comes from making it easier for customers to discover, buy, govern, and combine the capabilities of many third parties.
That matters because no software company will build every component of an end-to-end solution. As systems of record open to third-party agents and services, customers will want a simpler way to interact with the ecosystem. They will not want separate billing models, separate credit pools, and separate management systems for dozens of small tools.
This is where the orchestration layer becomes strategically important.
“The orchestration layer becomes more important than the system of record.”
The system of record may remain the data store in the background, but the orchestration layer can become the interface through which users interact, deploy agents, manage workflows, and understand the economics of what is happening. That layer could be an LLM interface, a functional application, or another form of agentic control plane.
The risk for systems of record is that they become invisible infrastructure: essential, but distant from the user experience and from the economic relationship with the customer.
The winners will be the platforms that combine trusted crediting, governance, third-party enablement, and an interface that makes complexity manageable.
“If you can command the tariff on every credit going through your system, you own the largest arcade in the world.”
That may be one of the defining commercial moves of the next decade: creating an ecosystem in which a common credit enables third parties to participate, customers gain simplicity and control, and the platform captures value from the activity that moves through it.