Manny Medina, previously best known as the founder of the sales automation startup Outreach, has impressed investors with his new venture, Paid. The young startup just closed an oversubscribed seed round of 21.6 million dollars, which was led by Lightspeed. Following a pre-seed round of 10 million euros raised in March, the London-based company has now raised a total of 33.3 million dollars and has not yet reached its Series A funding stage. A source familiar with the deal indicates the startup’s valuation is already over 100 million dollars.
Paid emerged from stealth mode in March with a unique proposition for the world of AI agents. The company does not create the agents themselves. Instead, it provides a way for the creators of these agents to charge their customers based on the value the algorithms deliver. This approach, often called results-based billing, is a growing theme in artificial intelligence.
Paid promises to help agent makers begin charging for the points of margin their customers save. This represents a new software pricing model for the AI age, moving away from the per-user fees common in the SaaS era or the one-time purchase fees from the client-server era. Per-user fees are problematic because agent makers must pay usage fees to both model providers and cloud providers. An unlimited use model could easily drive them into financial losses, an issue that also affects the vibe coding startup world.
Agent providers need to demonstrate the value their agents are delivering, especially since these agents often operate in the background. If the agents perform as intended, they will be assigned increasingly more work, and this growing workload may go unnoticed. As Medina states, if you are a quiet agent, you do not get paid. You need an infrastructure that allows the agent to charge for the additional work it performs.
However, charging a monthly fee for a limited number of credits, similar to the model used by AI model makers and vibe coders, is also risky for agent creators. Companies are reluctant to pay for low-quality AI output, which is still a common result. After billions spent on AI pilot projects, a recent study found that about ninety-five percent of enterprise projects delivered no value, with only five percent being put into actual production. Companies do not want to pay for agents that simply produce more unread emails.
One of Paid’s early customers is Artisan, a viral sales automation startup. Paid is also beginning to see success with SaaS companies that are exploring agents for their next phase of growth. The startup recently announced it has landed the ERP vendor IFS as a new customer.
Alexander Schmitt from Lightspeed noted that his venture firm has invested more than two and a half billion dollars into AI infrastructure and application companies over the last three years. During that time, they have observed that most AI pilot projects fail. He believes the core of the problem is that no one can effectively attach value to what agents are accomplishing today.
Schmitt considers Paid’s approach to be unique so far, stating it is something they have not seen anyone else build. He anticipates more competition will emerge in the area of results-based billing for AI agents if this model proves successful in helping agents enter the workforce at a large scale. New investor FUSE and existing investor EQT Ventures also participated in this funding round.