Tessa Lynch

Went through a few companies' software subscriptions recently and found the same pattern every time: multiple tools quietly doing the same job, nobody having noticed. How does your team currently catch that kind of overlap a regular review, a specific tool, or mostly just luck?

I started thinking about this after noticing a change in how some users interact with AI products. They don t want to open another app every day and repeat the same commands. They d rather set an intent once, let the agent keep working, and only hear back when something changes or needs their attention. That made me wonder if agent-native is less about the interface, and more about the task itself. A task feels more agent-native when: 1. context needs to persist over time 2. users care more about the outcome than the steps 3. the agent can decide when to act 4. human attention is only needed at key moments Maybe the real shift is from software we operate to responsibilities we delegate. For builders working on agents: what kinds of tasks have you found actually benefit from this model and which ones still work better as traditional software?