Wallaby has launched as a curated inference layer for open-weight models, starting with Kimi K3. The company emphasizes building durable value around models through infrastructure like agents, context, and memory. They are productizing their internal agent memory system, which stores memories as files with receipts, size thresholds, and age-out policies, alongside a secrets scan. This system is detailed in their public rulebook, wallaby-agent-rules, and their research log. AI
IMPACT Provides infrastructure for agent memory, addressing a key limitation in current AI agent capabilities.
RANK_REASON Product launch of an inference layer and agent memory system.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →