Pricing AI Agents: Seats, Usage, or Outcomes

Enroll

Agents break pricing in two ways at once. Cost scales with every step the agent takes, and value stops tracking seats, because an agent that replaces a workflow is worth the workflow rather than a license. This course works through the pricing questions that are specific to agentic products: whether to meter steps or outcomes, what happens to margin under multi-call chains, how per-task pricing behaves, and what buyers will actually sign. It pairs with the monetization course and also stands on its own.

The outcome

How does this prepare you for a job?

Every agent startup and platform team is arguing about this right now, so arriving with a framework is a real differentiator.

Designed pricing for agentic products across per-task, per-outcome and hybrid consumption models, each with a margin floor.
Modeled cost per completed task across a multi-step agent run, counting planning, tool calls and retries, rather than measuring per request.
Set usage meters and limits so that what the customer values moves in step with what they spend on tokens.
Priced pilots and enterprise rollouts for agent deployments, covering success criteria, caps and true-up mechanics.
Benchmarked competitor agent pricing, comparing per-seat copilots against per-resolution agents, and positioned against it.

Tools

AI tools learnt

The same instrument as all pricing work: the model, fed with real run data.

Hands-on skills

Model what an agent run truly costs, counting steps, retries, tool calls and the growth of the context window
Choose the meter, which is task, outcome, resolution, or hour of work replaced
Set a margin floor that still holds under the customer's heaviest workflow
Design caps, overages and true-ups that a buyer will accept
Price the pilot so it proves value without giving the product away
Read a competitor's agent pricing for what it reveals about their costs

Syllabus

Modules

Level: Basic

Why agent pricing is different

Seat value collapses, because the agent replaces the workflow rather than the user
Cost-per-task math set against cost-per-request math
The three meter families: usage, task and outcome

Cost model first

Instrument a real agent's runs and measure tokens per completed task
Variance, and the tax that retries impose
Margin floors derived from the heaviest workflow you have observed

Pick and defend a model

Match the meter to the way the customer counts value
A pricing one-pager for a real agent product
Stress-test it against three customer archetypes

Level: Advanced

Outcome pricing without going broke

Defining what counts as done, which brings verification costs and the risk of gaming
The conditions under which outcome pricing becomes a trap
Hybrids, meaning a base charge plus outcome kickers that still hold margin

Enterprise agent deals

Pilots, caps, true-ups, and what procurement actually does
The CFO conversation, framed as cost of labor, and where that framing breaks down
Success criteria that trigger expansion rather than disputes

Case studies from the frontier

Per-resolution support agents, per-seat copilots and per-token APIs, and what each meter implies about the provider's costs
What procurement at a large company does to each model
Defend the model you chose in a mock pricing review

Students can take the full course (both levels) or either level on its own.

Register

Waitlist

Join the course waitlist

No cohorts are scheduled yet. Join the waitlist and we’ll email you when registration opens.