Back

Why I'm Building Fluenty: Chat and Voice with Real AI

Your website's chat widget is dead. Traditional chatbots are glorified forms. I'm building something different.

Why I’m Building Fluenty

Your website’s chat widget is dead. Traditional chatbots are glorified forms. I’m building something different.

The Problem I Saw

A few months ago I was evaluating chat widgets for a project. The experience was… depressing.

  • Decision trees disguised as conversation
  • ”Select an option: 1, 2, 3”
  • Zero context between messages
  • User ends up looking for the support email
  • Natural conversations that understand context
  • Instant, relevant answers
  • Memory of previous interactions
  • Voice when typing isn’t enough

The “AI-Powered” New Ones?

Even the “modern” solutions have problems:

  • GPT wrappers with no business context
  • Generic responses that don’t know your product
  • Brutal latency (5-10 seconds per response)
  • Per-token pricing that scales horribly

Why? Because integrating voice with AI is complex. You need:

  • WebRTC for the call
  • Real-time speech-to-text
  • Intent processing
  • Text-to-speech response
  • All under 500ms to feel natural

Most companies don’t have the team to build this. Enterprise solutions cost $10k+/month.

What I’m Building

Fluenty is a chat and voice widget with real AI that any site can integrate with a single line of code.

Chat That Understands

Not a GPT wrapper. Deep integration with Retell AI. Trained business context, streaming responses, persistent history, native multi-language.

Real Voice

Actual WebRTC. Not “we record your audio and respond in 10 seconds”. Real-time voice calls with AI that responds like a human.

Built for Agencies

Multi-tenant from day 1. One account, multiple clients. Custom widgets per client. Integrated Stripe billing.

Full Analytics

Automatic transcription and recording. Per-call analytics. Organization-level insights.

Why Now

2011
WebRTC Launch
<500ms
Target Latency
$10k+
Enterprise Cost
99%
Text-Only Widgets

The Voice Moment

Voice-first isn’t the future. It’s the present.

  • AirPods everywhere
  • Alexa, Siri, Google Assistant normalized talking to machines
  • Users expect to speak, not just type

Mature Conversational AI

Two years ago, voice bots sounded like 90s robots. Today:

  • Text-to-speech indistinguishable from humans
  • Sub-second latencies
  • Models that understand real context

What It’s Not

Fluenty is not:

  • A replacement for human support
  • A bot that “solves everything”
  • The $50k enterprise solution

It’s a tool to:

  • Capture leads 24/7 with real conversations
  • Give instant answers about your product
  • Scale first contact without hiring more people
  • Offer voice when text isn’t enough

Current Status

API v2.7 stable

Battle-tested with a real pilot: a 360° redesign for Lomas Hospitality’s destination weddings vertical in Riviera Maya.

Analytics, billing, multi-tenant

Full agency support with Stripe integration.

Widget v0.5 in testing

Live on the Lomas Hospitality pilot handling real user conversations.

Public Beta

Coming soon with full documentation.

Coming Soon

In the next few weeks I’ll share more about:

  • Tech Stack: Fastify + Vue + Retell AI
  • Metrics from the first pilots
  • Pricing and business model
  • Beta access

If you’re interested in the conversational AI space, or you’re looking for something better than the same old chatbots, stay tuned.

Have a specific use case? An agency that needs this? Let me know at jlopezlira@gmail.com

JJ