The Typelessity Journal

Field notes on conversational AI booking.

Architecture decisions, latency math, compliance contours, and category essays from building the AI conversational booking widget.

13writeups
18,865words
82min total read
Featured · Comparison

Best AI booking widget for beauty salons and spas 2026: comparison and decision guide

Beauty booking is service-bundled, stylist-preference-driven, and walk-in-tolerant — the procurement requirements differ from medical or transfer. An honest comparison of the booking tools salons evaluate in 2026, with the dimensions that move a salon-owner decision.

10 min readAlex Isa
Read the comparison
02

Engineering

5 pieces
  1. Latency budgets for conversational AI booking: how to stay under one second

    GPT calls, enrichment APIs, render time — added together naively, that is three seconds. Typelessity targets a 1-second p95 user-perceived round-trip. Here is the budget, what stays inside, what gets cut, and what is honestly outside the budget.

    5 min1,144 wordsperformancelatency
  2. Whisper vs Web Speech API: why Typelessity uses Whisper for voice booking

    The Web Speech API is built into every Chromium browser and free at the point of use. Typelessity still uses OpenAI Whisper for booking voice input. Here is the accuracy, vocabulary-biasing, and confidence-scoring math behind that choice.

    5 min1,116 wordsvoicewhisper
  3. Cascade-aware corrections: how to handle field dependencies in conversational booking

    When a user changes one field mid-booking, every dependent field must be invalidated. Here is the dependency-graph algorithm Typelessity uses, the twelve lines of code behind it, and the UX rule that makes it readable to users.

    4 min971 wordsbugsalgorithms
  4. 25 languages, one prompt: how config-driven extraction beats per-language code

    Most multilingual chatbots ship a separate codepath per language. Typelessity ships one prompt template, one config, and 25+ languages work the same day. Here is the architecture and the 80 strings of localization that remain.

    4 min949 wordsi18nprompts
  5. Why we replaced the booking form with a single GPT call

    Single-call extraction replaces the multi-step wizard: one model round-trip pulls every field from one user sentence. Here is the architecture, the prompt, the anti-hallucination guard, and the latency budget that justify it.

    4 min1,001 wordsarchitecturegpt-4.1-nano