Concept detail
Agent autonomy
0 tagged videos · 5 connection videos · degree 0
Publication timeline
- 2026-061 video
- 2026-082 videos
- 2026-092 videos
Source videos
Observed neighbors
Context size
The more autonomous an agent is, the more competing tool descriptions it must hold in context, increasing the chance of poor tool selection.
1 source videos
Building Agentic Systems with Python, LangGraph, MCP, and A2A [PyCon DE & PyData 2026]End-state expression
Mutually supporting: specifying an end-state only works if the model is trusted to choose the intermediate planning and execution steps.
1 source videos
How Anthropic Builds: Lessons from Labs — Mike Krieger, Anthropicloop engineering
Loops are the mechanism that converts interactive prompting into background autonomous operation.
1 source videos
How Anthropic Builds And How Engineering Will Change Soon | Thariq Shihiparoutcome-based metrics
Measuring user autonomy requires outcome ground truth, which is the same evidence needed to evaluate whether a harness was absorbed successfully.
1 source videos
Google DeepMind's Logan Kilpatrick: Why the Model Eats the Harnessverification frameworks
As agents generate more code and operate with more autonomy, verification capacity becomes the gating constraint on safe deployment rather than a supporting tool.
1 source videos
OpenAI researcher on agent swarms & recursive self-improvementRelated tagged insights
No core insights are tagged with this concept.