inioluwa.
🤗 Hugging Face
All projects

Agents · Public Hugging Face Spaces

Voice-agent workflow labs

Make an agent’s decisions replayable.

Repository
Eunini/huggingface-portfolio
Stack
TypeScript · State machines · Tool traces · Hugging Face
Headline
34 public regression scenarios

Two public, inspectable simulators for voice-agent workflows: an AI SDR and a hotel receptionist, backed by a scenario dataset.

AI SDR workflow labHotel receptionist labScenario dataset

01The problem

A voice agent that books, follows up or takes payment makes decisions that need to be checked, not just heard. Those decisions should be replayable as tests.

02The approach

  1. Model permission, opt-out, scheduling, consented follow-up and interrupted audio turns as explicit events with expected outcomes.
  2. Model priced offers, expiring inventory holds, exact verified-payment events and safe cancellation boundaries for a fictional hotel.
  3. Publish the 34 scenarios as a dataset with executable reference engines so the same cases can regress a real system.

03The evidence

  • Both labs run in the browser as static Hugging Face Spaces with visible tool traces.
  • The scenario dataset and reference engines are public.

04Scope & status

Deterministic companions to private projects. They do not run an LLM or telephony provider, send messages, take money or create real bookings; provider and identity checks are stubs.

Public source is available for independent inspection.

Next projectGridSense