The End of Paying for Time – AI Is Making Outcome-Based Pay Inevitable

01 /The Concept

Companies pay out roughly $50 trillion in wages every year. And yet most payroll systems can only track one thing: time. Hours worked, days logged. Anyone who wants to pay for results instead is stuck maintaining manual spreadsheets or cobbling together custom tooling just to automate what should be straightforward.

Grade's founders ran into this problem firsthand while promoting their own apps through influencers. After a while, they wanted to move away from flat per-placement fees and pay for actual reach and signups instead – only to discover there was no software that could handle it. They ended up tracking everything by hand.

That experience became Grade: a platform for managing performance-based compensation. The current focus is influencer payouts, though the company is explicit that this is just the starting point.

Onboarding an influencer takes nothing more than entering an email address. The platform sends an invitation link and walks the creator through the entire setup process – connecting accounts for tracking impressions and clicks, submitting payment details, and everything in between.

For each influencer or campaign, brands configure the payment structure: flat fee per placement, performance-based pay, or a base plus performance bonus.

Grade pulls the relevant data automatically from connected accounts and calculates each influencer's payout according to the rules set for them.

Payments can run on any cadence – daily, weekly, or monthly. A useful feature: brands transfer a single lump sum to Grade, which then distributes individual payouts to each creator's account, converting currency as needed across 190 countries.

Grade is currently in Y Combinator. The platform has been operating for several months already; over the last 30 days, it processed $380,000 in payouts – a 120% increase over the prior 30-day period.

02 /Why It Matters

Paying people for outcomes rather than hours has been a recognized goal for decades. But right now it's become urgently relevant for a surprisingly modern reason.

That reason is AI agents.

AI agents can, in principle, perform the same work as employees and contractors. But when companies try to deploy them, a structural problem emerges: how do you actually pay for an AI agent?

Traditional SaaS pricing models don't fit.

A flat monthly subscription doesn't work because an AI agent's costs are highly variable – every call to an underlying model like GPT or Claude is a real expense. A fixed subscription price that's too low destroys the developer's margin; one that's too high scares away customers. There's no sweet spot.

Usage-based pricing might seem like the solution. But it creates a different problem – for the customer. An AI agent could rack up enormous model costs, consume a significant budget, and still deliver garbage output.

With human employees, the risk is similar in nature but bounded by salary. With AI agents, there's no natural cap. The bill for a failed agent run could easily exceed what a human doing the same job would have cost by several multiples.

Process of elimination leads to the obvious conclusion: AI agents should be paid for results. This also happens to align developer and customer incentives perfectly. The customer knows exactly what they're paying for and can evaluate whether the cost is justified. The developer is motivated to maximize outcomes while minimizing model token consumption – which directly improves margins.

But for that to work, developers need tooling that can measure, calculate, and bill against outcomes. Paid ([related review](/review/skoro-vse-budut-rabotat-tolko-po-takoj-biznes-modeli)) is one of the startups building exactly that – raising a €10M first round last year, followed by a $21.6M second round shortly after.

Noteworthy: Paid was founded by the founder of Outreach, who built that company to $250M in ARR and a $4.4B valuation before stepping down to launch Paid. When someone leaves a company at that scale to start something new, they're betting on the new thing being bigger.

But here's the deeper implication. If companies are already building the infrastructure to measure and pay for AI agent output – why not apply the same framework to human workers?

If you can compare the cost of getting a result from an AI agent versus a human, you suddenly have a rational basis for deciding which to use. But to make that comparison, you need to know what a human-delivered result actually costs. And once you know that – it's a short step to paying humans by the result too.

03 /Opportunities

The chain of logic here runs like this: AI agents require outcome-based pay → outcome-based pay requires measuring results → measuring AI agent results requires measuring human results too → which will inevitably lead to humans being paid by outcomes as well.

For a long time, performance-based pay was a nice-to-have – aspirational but avoidable. Now it's becoming a hard requirement. Because deploying AI – which practically demands it – is itself becoming a hard requirement. And once the infrastructure for tracking outcomes is in place for AI, there's no logical reason to exclude humans from the same system.

The broad trend: every company will eventually move to performance-based compensation. The infrastructure to make this possible – platforms that automatically track, calculate, and execute payouts – is what needs to be built.

The most compelling build here is probably not a tool that separates human from AI workers. In many workflows, the meaningful result only appears at the end of a chain of actions performed by a mix of people and agents. A platform that tracks outcomes across that entire chain, regardless of who or what performed each step, is more valuable than one that siloes the two.

Today's Grade is the simplest possible version of this idea. There's a lot of runway ahead. Which, honestly, is the best time to enter a space – get in now, establish a foothold, and expand from there.