Bigspin Blog

Closing the gap between evaluations and system improvements
Evals

Closing the gap between evaluations and system improvements

Now that even the humblest prompt is an “agent”, we can focus on what really matters
Agents

Now that even the humblest prompt is an “agent”, we can focus on what really matters

Superforecasting the future of software development
AI Forecasting

Superforecasting the future of software development

Claude 4.1, Genie 3, and GPT-5 are compound AI systems, not simply models
LLM models

Claude 4.1, Genie 3, and GPT-5 are compound AI systems, not simply models

Prompting tricks are dead, but prompts matter more than ever before
Prompt engineering

Prompting tricks are dead, but prompts matter more than ever before

The soul of a new LLM
LLM models

The soul of a new LLM