Quick answer
When a business wants a production‑ready voice agent, the first choice is which Voxovo 1.0 pipeline to use. Speed, Smart, and Ultra modes give you a spectrum from minimal turn latency to maximum linguistic fidelity, and knowing your use case is the key to picking the right one.
Voxovo 1.0 is a fully managed stack that bundles speech‑to‑text (STT), large‑language‑model (LLM), text‑to‑speech (TTS), and voice synthesis into a single, low‑latency pipeline. The stack is pre‑validated, so you never need to juggle individual APIs or tune VAD thresholds yourself.
Speed mode (`pipeline_mode: "voxovo1"`) is the lightest tier. It keeps turn latency under 500 ms and uses fixed, high‑performance models for STT, LLM, and TTS. Because it’s optimized for speed, it is ideal for high‑volume outbound qualification or first‑touch support where the user expects a rapid response.
How Voxovo AI works on Voxovo
Smart mode (`pipeline_mode: "voxovo1_smart"`) adds an accuracy layer. The same managed stack is used, but the LLM and TTS models are larger and fine‑tuned on conversational data, pushing turn latency to around 800 ms. Smart is best for scenarios that demand correct intent detection and nuanced language, such as complex billing conversations.
Ultra mode (`pipeline_mode: "voxovo0_ultra"`) is the most expressive tier. It swaps the standard LLM for Gemini Live, enabling real‑time speech‑to‑speech. Turn latency remains comparable to Smart, while the system can handle multi‑turn context and incorporate on‑the‑fly knowledge base snippets, making it perfect for advisory or high‑value sales calls.
Choosing between the modes comes down to a trade‑off matrix: latency vs. accuracy vs. feature set. If your pipeline must handle 1,000 calls per minute with no room for buffering, Speed wins. If the conversation requires deeper reasoning or multilingual support, Smart or Ultra are more appropriate.
The Presence Buffer is a Voxovo hallmark. By pre‑caching filler audio, the system guarantees callers never hear dead air while the LLM or tool chain is busy. This subtle benefit is often the difference between a smooth experience and a frustrated caller, especially in Speed mode where latency is minimal.
Implementation checklist
Live Sheet and Live Calendar allow agents to pull in first‑party data or book meetings during a call without leaving the conversation. The zero‑setup integration means a sales rep can pull a prospect’s recent interactions from Sheets or schedule a follow‑up directly from the voice interface, boosting productivity across all pipeline modes.