Named agents that watch one domain each, ground in its 360, and recommend the next action — the reasoning layer over the mesh.
Each agent owns one job — collect cash, defend renewals, capture cross-sell, finish integration, watch credit, hold pricing, fill capacity. They read the same governed data the 360s show, surface the live finding, and recommend the move. This is Doc 7 of the stack: reasoning over the graph.
Every card reads the same governed data its 360 shows — a live finding, the recommended move, and the value at stake.
Finding. $590K is >60 days late; the slowest-paying units (Gurgaon Tech Park 67d, CA Real Estate Advisory 66d) drive it.
Recommends. Run a collections sprint on the aged book — frees ~$-15912K toward the 48-day target.
Finding. $9M of cross-sell opportunity; Gupta Enterprises alone has $100K untapped.
Recommends. Attach a 2nd/3rd service (lead with monitoring) on the top single-service accounts.
Finding. $860K of ARR is at-risk across the next four quarters (churn-flagged or contraction-likely).
Recommends. Trigger save-plays on at-risk renewals; each churn point on the $314M base is $3.1M of ARR.
Finding. $5M of acquired revenue still integrating; CA Real Estate Advisory is the laggard at 45%.
Recommends. Sequence CA first on a dated 90-day plan — unlocks margin, cash and office-grain actuals.
Finding. 1 account(s) at High churn/credit risk (e.g. Chaudhary Family); several customers carry conflicting payment terms across brand systems.
Recommends. Move distressed accounts to credit hold; reconcile term conflicts on the golden record.
Finding. Standalone tools discount at 16% vs 10% on governed CPQ — straight margin leakage.
Recommends. Migrate Arora PMS-2024/CA Investment Banking/MCEF-Alpha-Gurgaon Tech Park quotes onto Client Associates CPQ with one price book and approvals.
Finding. Bengaluru runs lowest at 68% productive time with 22 open reqs.
Recommends. Move acquired crews onto common dispatch; lift utilization toward 78% — capacity already paid for.
Grounded in the live cockpit snapshot via Azure OpenAI.