Definition
Agentic coding is AI-driven software development where a model plans, edits files, runs commands, and iterates on errors largely on its own, rather than acting as a turn-by-turn autocomplete. The promise is end-to-end task completion; the open problems are reliability over long sessions, repository-scale context, and trust.
Episodes covering this
Worth reading next
Papers we haven't done a deep dive on yet, but would recommend on this topic.
- SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering
- AutoCodeRover: Autonomous Program Improvement
- SWE-bench: Can Language Models Resolve Real-World GitHub Issues?
- FunSearch: Making new discoveries in mathematics using large language models
- OpenEvolve: Open-Source Implementation of AlphaEvolve
- Focal Loss for Dense Object Detection
- FunSearch: Making New Discoveries in Mathematical Sciences Using Large Language Models
- Draft, Sketch, and Prove: Guiding Formal Theorem Provers with Informal Proofs
- Scaling LLM Test-Time Compute Locally Increases the Concentration of the Hardest Problems
- Agent-as-a-Judge: Evaluate Agents with Agents
- Verified Multi-Step Synthesis using Large Language Models and Monte Carlo Tree Search
- RLEF: Grounding Code LLMs in Execution Feedback with Reinforcement Learning
- Scaling LLM Test-Time Compute Optimally Can be More Effective than Scaling Model Parameters
- Voyager: An Open-Ended Embodied Agent with Large Language Models
- Code as Policies: Language Model Programs for Embodied Control
- Self-Refine: Iterative Refinement with Self-Feedback
- Reflexion: Language Agents with Verbal Reinforcement Learning
- Agentless: Demystifying LLM-based Software Engineering Agents
- Autoformalization with Large Language Models