Quinn Slack(@sqs):This is exactly what we want to achieve with Amp: get some simple primitives right (like threads and orbs and runners), and let the models use them. No tricks. No "must-have" skill stacks. And it means Amp gets better when the models get better.

2026.09.18 07:56

This is exactly what we want to achieve with Amp: get some simple primitives right (like threads and orbs and runners), and let the models use them. No tricks. No "must-have" skill stacks. And it means Amp gets better when the models get better.

rafael coutinho@c0utinn

2026.09.18 01:11

using amp just feels right, finally the broken idea of having a bunch of agents to simulate productivity feels actually useful. i've tried different orchestrators and all that shit, but they all lose to amp just by how simple it is. having different orbs that are ephemeral but easy to use (reminds me of the honeymoon with nix) with clankers is just delicious. i've been using clankers since gpt2 and this is the first time something really feels useful to me. great work, especially with how consistent and easy it is to set up orbs @thorstenball awesome!!!

108

轉發到社區

熱門用戶