gepa-ai
Automatically optimize prompts, code, and agent architectures using LLM-based reflection, achieving frontier performance with 90x lower costs than Reinforcement Learning.
At a Glance
- AI Engineers
- Enterprise AI Teams
- LLM Researchers
- Open Source Community
AI Tools by gepa-ai
(1)
GEPA
Genetic Prompt Auto Optimizer
Discussions
No discussions yet
Be the first to start a discussion about gepa-ai
Latest News
Learning, Fast and Slow: Towards LLMs That Adapt Continually
Scaling GEPA with Combee: Parallel Prompt Learning for Self-Improving Agents
Confidence-Aware Prompt Optimization for LLM Classification
Automatically Learning Skills for Coding Agents
Products & Services
A framework for optimizing any system with textual parameters using LLM-based reflection and Pareto-efficient evolutionary search.
A universal API for optimizing prompts, code, agent architectures, and configurations against any evaluation metric.
Market Position
Outperforms GRPO (Reinforcement Learning) on accuracy while being 90x cheaper and 35x faster, offering fully interpretable and human-readable optimization traces.
Leadership
Founders
Lakshya A Agrawal
PhD student at UC Berkeley Sky Lab, Co-founder of Bespoke Labs. Lead developer of dspy.GRPO and GEPA. Previously at Ai2.
Shangyin Tan
PhD student at UC Berkeley Sky Lab working on Compound AI systems and agents. Previously worked at Databricks.
Executive Team
Lakshya A Agrawal
Lead Researcher/Developer
PhD Student at UC Berkeley, expert in prompt optimization and DSPy.
Shangyin Tan
Lead Researcher/Developer
PhD Student at UC Berkeley, expert in agentic AI and compilers.
Board of Directors
Founding Story
GEPA was developed at the UC Berkeley Sky Computing Lab to overcome the limitations of Reinforcement Learning for prompt optimization, specifically focusing on data efficiency, cost, and interpretability through natural language reflection.
Business Model
Revenue Model
Open Source (Free). Users pay for their own LLM API costs (OpenAI, Anthropic, etc.).
Pricing Tiers
Available via pip install gepa. Free to use with your own API keys.
Target Markets
- AI Engineers
- Enterprise AI Teams
- LLM Researchers
- Open Source Community
- Agent prompt optimization
- Code generation tuning
- Cloud scheduling policy discovery
- Reasoning task performance improvement
- Cost reduction for enterprise AI agents
- Shopify
- OpenAI
- Databricks
- Pydantic