Selection
Click any node to see what it connects to, what feeds it, and how data flows through the harness.
The harness flow
Input stage
Qualify transcripts, session context, and system prompts are assembled and fed into the agent's working memory alongside long-term memory stores.
Reasoning stage
The LLM Agent drafts the staffing ledger using agentic tools for scope math, segregation checks, and gap detection.
Output stage
A human reviews and confirms the draft before it produces two views: the client SOW and the internal Staffing Briefing.
Learning loop
When an engagement closes, delivery data flows through a Summarizer that distills reconciliation data back into procedural memory, making the next staffing prediction more accurate.