Eric Siu breaks down Cognition's $10 million AI guarantee for Devin, which refunds customers if the tool doesn't deliver engineering value worth what they paid. The guarantee shifts focus from token usage to actual productive output, measured by comparing Devin's work to human engineer time on real enterprise code.
There's a new AI productivity guarantee, and I think we're going to see a lot more of this. So I'm pulling this up right now. So Cognition says AI should earn its keep. Introducing the AI productivity guarantee. If Devin, which is their product, delivers less engineering value than you're paying for, Cognition will fund your usage until it does, up to $10 million. Okay, so that's a good guarantee, man. That's a good offer. So it's time the AI industry— for the AI industry to stop maximizing tokens and start maximizing productive output, which is what we've been saying for a while. Like, it's not just about token maxing, it's about output maxing, right? So we build a system to estimate whether an agent's output was actually productive, and if so, how long a human engineer would have taken to do the same work, validated on real engineers' time estimates working in enterprise code bases. So they do a blog post on this, but I think you gotta go in this direction if you want to stay alive, because these, these other open source models are catching up too quickly.
