A user says AI models already do more than they can use for coding
A reply suggests giving agents broader tasks, involving them earlier and providing what they need to verify their work—then allowing them to merge code independently once confidence grows.
TLDR
One user says existing models are already more than good enough for their coding work, leaving them unexcited about Fable 5.1 or Astra while they learn how much to trust their current tools. A reply recommends broader prompts, earlier agent involvement and asking agents to work longer, with what they need to verify their work. Once confidence grows, it suggests letting the agent merge code on its own.
Combined views
—
1 Source, first seen ago
