— Case study

An AI avatar trained on your tutor's own lessons — not a generic one standing in for them.

BePolyglot is a language-learning marketplace built around a simple frustration with existing apps: lessons are too rare to build a habit, and self-study apps never let you actually talk to anyone. Feneco is building the product and the platform behind it — three ways to train, one record of progress that follows the learner across all of them.

Product
Language-learning marketplace
Training modes
Human, Hybrid, AI Avatar
Roadmap
6 phases, 12 months
Status
In development

Three modes, one progress file

Most language apps force a choice: book a human tutor, or practice alone with a chatbot. BePolyglot treats that as a false choice.

Human

A native tutor leads the lesson. Full control, full nuance, scheduled sessions.

Hybrid

The tutor sets the strategy live; their own AI twin drills it with the student between sessions.

AI Avatar

A 100% AI teacher, available on demand, adaptive difficulty on every turn.

All three write to the same progress record — streak, fluency score, minutes spoken, active vocabulary. Switching modes never resets it.

The avatar is the tutor's, not a substitute for one

In hybrid mode, the AI isn't a separate teacher competing with the human one — it's that tutor's own digital twin, trained specifically on their lesson plans and teaching voice. The tutor sets the strategy in a live lesson; between sessions, the student drills the same material with the same method and unlimited repetitions, no scheduling required. Each tutor manages their own twin from a dedicated training panel in their dashboard — upload lesson materials, review processing status, test the twin themselves before a student ever talks to it.

Correction that doesn't interrupt

Live lessons run against a real-time transcript. Mistakes get flagged inline as the conversation happens, instead of an assistant breaking in mid-sentence — feedback without breaking the one thing that actually builds fluency: talking.

What's built, and what's next

The product runs on a six-phase, twelve-month roadmap, foundation first: the marketplace and learning experience before the AI layer. What exists today is a full authenticated product — separate teacher and student dashboards, onboarding, curriculum management, live classroom sessions, messaging, billing, and the tutor's AI-twin training panel — built on Auth0-backed authentication. The avatar's actual conversational and voice layer is the next phase of the roadmap, built on top of that foundation rather than bolted onto it after the fact.

Status

In development — no public metrics yet. This page will update as the platform ships.

If you're weighing how much of a product to build before the AI layer earns its place, let's talk.