AskAlf drops self-hosted autonomous agent workforce

AskAlf shifts the focus from simple task agents to a persistent, self-healing "workforce" that manages its own knowledge gaps.

• –The 10-layer memory architecture (stored in pgvector) enables procedural learning, allowing the system to optimize repetitive tasks like security scans over time
• –Support for the Model Context Protocol (MCP) ensures it can integrate deeply with existing developer toolchains and local data sources
• –While currently reliant on Claude and OpenAI, the commitment to Ollama/vLLM support makes it a strong contender for fully air-gapped enterprise AI
• –The "investigation ticket" system mimics human project management, creating a more structured and traceable autonomous execution loop than raw agent swarms