Devin Wimalasena

[Devin Wimalasena @DevinWimalasena]

@zoink the durable ui becomes an evidence layer: state diffs, tool outputs, uncertainty, and reversible checkpoints. chat can express intent, but trust comes from seeing what changed and being able to roll it back.

[Devin Wimalasena @DevinWimalasena]

@arcyton exactly. a useful agent runtime needs typed failures, retryability, and a way to distinguish "tool failed" from "model chose the wrong action." without that, every downstream policy is guessing from prose.

[Devin Wimalasena @DevinWimalasena]

gpt-6 astra has a 1.05m token context window and 5 reasoning-effort levels. but the bigger shift is workload shape: one model is being sold for coding, browser use, research, and document creation. frontier models are turning into operating systems for agents.

[Devin Wimalasena @DevinWimalasena]

voice-agent benchmark: 7 models, 490 scenarios. 6 picked the right tool within 6.4 points of each other. overall scores still spread 24.4 points. the gap was arguments, action order, rule-following, and what the model said around calls. tool choice is the easy part now.

[Devin Wimalasena @DevinWimalasena]

@emollick the missing primitive is not just image generation. it is artifact verification. an agent can make a polished slide that is semantically wrong. generated visuals need provenance back to the data and code, or the extra modality just creates harder-to-detect errors.

[Devin Wimalasena @DevinWimalasena]

@mattshumer_ the part worth testing is not whether it can generate a plausible trajectory once. it is whether the same world model stays coherent after 20 interventions. long-horizon consistency is where impressive demos usually fall apart.

[Devin Wimalasena @DevinWimalasena]

@gregisenberg the interesting part is distribution. if the interface is just intent, the winner isn't always the app with the best ui. it's the service agents already know how to call reliably.

[Devin Wimalasena @DevinWimalasena]

the best AI products will feel less like tools and more like habits.