Outcome-Based Billing Built for AI Agent Companies

01 /The Concept

Skope is building a billing platform designed specifically for AI companies – one that goes beyond legacy subscription and usage-based models to support outcome-based pricing.

With Skope, a developer can assign a distinct billing model to each AI agent – choosing a single model or blending several. The most compelling option is, of course, charging for results: the logical next frontier for monetizing AI agents.

The platform comes pre-loaded with billing templates for the most common outcomes an AI agent might deliver – a qualified lead generated, a meeting booked, an email sent, a support ticket closed, a contract signed, and more. Integration is via API: in the simplest case, a completed action (say, sending an email) triggers the billing event. In more complex scenarios, the agent pulls outcome data from the client's systems – for instance, reading closed deals from a CRM to trigger contract-based fees.

Skope is currently in the Y Combinator batch and only posted on the YC site recently, so public information is still sparse.

02 /Why It Matters

The biggest wins in startup-land go to companies that arrive at exactly the right moment – when a shift in technology or behavior creates demand that didn't exist before.

Skope's sole blog post so far is titled "Outcome-Based Pricing: Why Now?" – and the argument is worth unpacking.

Software started as a packaged product: you bought a box. Then the internet brought SaaS, and subscriptions became standard – priced by feature tier and seat count. Simple, but it produced a lot of shelfware: companies paying for unused seats. As cloud services grew more compute-intensive, pricing shifted to usage-based models, eliminating waste since idle resources cost nothing.

Now AI agents are reshaping the picture again. An AI agent doesn't just assist people – at its best, it completes tasks autonomously. And nobody pays a human employee purely for hours logged (the subscription analogy) or lines of code written (the usage analogy). People get paid – and bonused – for results.

The same logic should apply to AI agents. Which means outcome-based billing is the natural pricing model for this era, and platforms that support it are infrastructure for a new category.

That said, outcome-based pricing raises real questions:

- Clients are unaccustomed to monthly bills that swing widely depending on the agent's performance. - Developers struggle to forecast revenue when it's tied to client outcomes. - Defining what counts as a "successful outcome" – and what it's worth – is genuinely hard to negotiate. - There's currently no widely-adopted billing platform that handles all of this with sufficient flexibility.

But here's the twist: these objections only surface when business models on both sides are unclear or poorly defined.

- If a client can't articulate what a result is worth, they don't understand their own unit economics. - If a client won't pay market rate for that result, their business model doesn't work. - If a developer can't clearly state what their agent delivers and what it's worth, they haven't defined their value proposition. - If market-rate pricing for an outcome doesn't cover the developer's costs, their business model doesn't work either.

In other words, outcome-based billing forces both buyers and sellers to quantify their models precisely – which is a sign of progress, not a problem. This model is going to become standard. Companies that resist it will be exposed as running unviable businesses.

These same arguments are made explicitly by Paid ([covered here](/review/ty-malo-zarabatyvaesh-potomu-chto-dengi-ne-za-to-berjosh)) in its "AI Agent Developer Manifesto." The argument starts with the same premise – subscription and usage-based pricing no longer fit the AI era – and ends with the same conclusion: agents should earn on outcomes, not features.

Paid built a similarly flexible billing platform that handles subscriptions, usage, outcomes, and any combination. But it goes a step further: a separate module calculates the true cost of resources consumed by each agent task, giving developers a way to protect margins under any pricing model.

Paid raised its first €10M in March. Notably, its founder previously built Outreach to $250M in ARR and a $4.4B valuation – then stepped back from that company to launch Paid, apparently convinced the opportunity here is even larger.

03 /Opportunities

Outcome-based billing is clearly where this industry is heading. And whenever the market shifts from old to new, incumbents that can't adapt create an opening for new entrants.

But the real opportunity is bigger than just billing. Next-generation platforms here will need to go in two directions at once:

First, inward: they'll need to model the developer's own cost structure – functioning partly as financial analytics tools, not just payment processors. Paid has already started down this road.

Second, outward: they'll need to help developers prove ROI to clients – showing that what clients are paying for is genuinely worth it. Paid already has a module that, in its own words, "transforms amorphous AI value into measurable return on investment" – helping developers renew contracts and compete on metrics rather than features.

Compared to Paid, Skope looks lean right now. But the category is large enough for multiple winners, and Skope's presence in the current YC batch confirms that. The direction is clear: build billing infrastructure that doesn't just support outcome-based pricing, but helps both sides of the table build and validate the business models to make it work.