Lago a republié ceci
IF AI agents replace one or several FTEs (aka "full time equivalent humans"), how should they be priced? It's too soon to say for AI agents that are mostly priced "per seat" at the moment (i.e., a fixed price per user, regardless of the task the agent is performing, the value it brings, or the compute price it costs to run the agent). Copilots pricing might be a good proxy or at least data point to think about AI agents pricing, and I found Tomasz Tunguz analysis and charts the most useful. TLDR: The Variance between the AI versus Base price ratio is high, meaning... it "depends". Our take at Lago is that: 1️⃣ Pricing is as much science than art (humans choose products and react to pricing, and humans... aren't fully rational!), especially in such a "new" field (i.e., what agents can do today is nothing like what they'd be able to do in a few months). 2️⃣ Velocity is key: so the most important is to be able to iterate quickly on pricing, to make price match value, and anticipate/react to market shifts: cost of compute, demand, competition, just to name a few dimensions. 3️⃣ To offer pricing velocity and flexibility: we've built Lago to be developer first-, agnostic, and transparent (open-source). There's no big secret there: if you understand our code and how we built our API, and have a community of users around you, you're able to use it, build on top of it, and customize it faster and better. Once again we're both proud and humbled (weirdly enough you can combine both feelings 😅) to serve leading teams and companies, such as Mistral, Groq, Together, and many others! (*Copilot vs Agents: Agents are supposed to automate entire tasks versus "copilots" support "pilots/humans".) Source: https://lnkd.in/esN4uVBf)