NeFut Logo NeFut
Admin Login

[CS.AI] From Monolithic Blending to Agentic Orchestration: Dynamic Response for Conversational Assistants at Scale

Published at: 2026-09-10 22:00 Last updated: 2026-09-12 06:35
#AI #optimization #LLM

This paper reports a production migration of a customer‑support assistant at a large accommodation marketplace from a monolithic Qwen3-235B-A22B blended responder to a Dynamic Response (DR) architecture. The original system performed retrieval, action selection, escalation, and wording in a single model path, whereas the new system separates these roles:

The migration also altered prompts, alignment, and serving. To isolate architectural effects, the authors replayed identical turns and measured changes. Key findings:

A low‑ramp A/B test reproduces escalation reductions: hard‑escalation responses drop from 5.60% to 3.08% and soft‑escalation from 9.56% to 2.49%, while handoff volume stays roughly steady; self‑solve shows a directional increase of about 5.1 points (95% CI [-2, +12]).

Serving optimizations cut orchestrator P90 latency from 3.87 s to 2.24 s with GPU footprint reduced by roughly one‑third, and self‑hosting lowers estimated annual model‑serving cost by more than an order of magnitude.

Review

Original Source: https://arxiv.org/abs/2609.05758

[h] Back to Home