# Let Decision Models Guide Your Voice Assistant

DevFeed: [Let Decision Models Guide Your Voice Assistant](<https://devfeed.tech/articles/let-decision-models-guide-your-voice-assistant-117172.md>)

Original publisher: [Read original article](<https://towardsdatascience.com/let-decision-models-guide-your-voice-assistant/>)

Author: Shuai Guo

Published: 2026-11-04T12:30:00Z

Content type: tutorial

Language: en

Sources: [Towards Data Science](<https://devfeed.tech/sources/towards-data-science.md>)

Topics: [Conversational AI](<https://devfeed.tech/topics/conversational-ai.md>), [AI Models](<https://devfeed.tech/topics/ai-models.md>), [decision-making](<https://devfeed.tech/topics/decision-making.md>)

Tags: [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [case-study](<https://devfeed.tech/tags/case-study.md>), [decision-models](<https://devfeed.tech/tags/decision-models.md>), [gpt-live](<https://devfeed.tech/tags/gpt-live.md>), [jev](<https://devfeed.tech/tags/jev.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llm-agents](<https://devfeed.tech/tags/llm-agents.md>), [llm-applications](<https://devfeed.tech/tags/llm-applications.md>), [voice-assistant](<https://devfeed.tech/tags/voice-assistant.md>)

## AI overview

The example separates live speech from inspection decisions: GPT-Live listens, speaks, and transcribes both sides, while Microsoft Decision-1 evaluates the conversation and selects the next action. A Python app sends the accumulated transcript to Decision-1 in one request containing four requirement checks and a next-step choice. Each check returns `ready`, `issue`, or `needs_evidence`; the action can ask about a requirement, defer while a check is blocked, or finish once all requirements are assessable, even if one has a problem. The example shows evidence being corrected later: a bedding check is reopened, while a confirmed sink leak remains an issue until maintenance repairs it and a new test passes. A WebSocket carries the ongoing GPT-Live exchange and delegation events; asynchronous tasks let audio and transcript events continue during assessments. The post says the same pattern can use Jev or OpenAI's Decisions API, with the decision request changed. Its worked scenario is synthetic, and implementation details are partly in a separate notebook.

## Source excerpt

GPT-Live does the talking. A decision model tracks what's confirmed and picks what to ask next. The post Let Decision Models Guide Your Voice Assistant appeared first on Towards Data Science.