Definition
Plain language
Whether an AI agent can actually carry out the action it decided on, without typos or formatting mistakes breaking the call.
As stated in the literature
In Life-Harness's lifecycle taxonomy, the first stage of agent failure — when the agent's chosen action fails to execute correctly due to syntax, formatting, or tool-invocation errors.
Why it matters: Many agent failures aren't reasoning errors at all — they're basic formatting bugs that prevent any tool call from running, so isolating this stage helps target the real fix.
For example, if an agent decides to call a search tool but emits malformed JSON arguments, the call fails at the action realization stage before anything else can go wrong.
Heard on the show
“The action realization layer does this for the model's tool calls.”Episode 071 — When the Model Is Fine and the Plumbing Is Broken: Fixing Agents at the Interface